Differentiated Understanding

Επεισόδια

• 
  Pony.ai’s Founder and CEO James Peng on What It Takes to Scale Robotaxis 21.07.2026 55λ

  When James Peng founded Pony.ai in 2016, many in Silicon Valley believed autonomous driving was only three to five years away. But he expected it would take at least a decade, because the challenge was never just teaching a car to drive. Commercialization also required regulatory approval, public trust, reliable fleet operations, and a cost structure that could support large-scale deployment. Ten years later, his vision is becoming reality.In this conversation, we start with his founding journey, the milestones and how Pony.ai became a leader in the autonomous driving space. We also discuss the gap between assisted driving, Level 4 autonomy, and the longer-term goal of Level 5, as well as how Pony.ai uses simulation, real-world driving data, and increasingly capable AI models to improve safety. James explains that the hardest problems are often not the obvious ones but interpreting unpredictable human behavior and handling rare edge cases consistently.The conversation also explores China’s cost advantage in robotaxis. A mature automotive and electronics supply chain, close collaboration with automakers, and faster iteration can materially lower vehicle and system costs. But moving into new markets still requires Pony.ai to adapt to different road conditions, regulations and driving cultures, from trams and roundabouts to local pickup behavior.James’s broader point is that the industry has focused too heavily on the initial technological breakthrough. Getting a car to drive itself is only the beginning. The next phase is about deployment density, utilization, maintenance, charging, remote support, and economics. At the end of the conversation, I asked what he believes is underrated. James, an experienced operator, replied - scaling. Pony.ai may have crossed the zero-to-one threshold, but the harder task is scaling from one to ten, and eventually from ten to one hundred. Check out this insightful conversation.For more interesting conversations with people who are charting the way of the future of AI, check out the podcast tab or follow us on Spotify!Chapters02:36 Why James Peng founded Pony.ai06:36 The milestone that proved robotaxis could work09:27 How passengers learned to trust driverless cars11:41 Level 2, Level 4 and Level 5 autonomy18:59 How AI and simulation improve self-driving24:13 Teaching cars to understand human behavior29:11 China’s cost advantage and global competition35:03 Expanding robotaxis into international markets41:13 Why Pony.ai is also building autonomous trucks48:40 Adapting to new cities, roads and driving cultures53:45 Why scaling is often harder than reaching zero to oneTranscriptGrace Shao: Hi everyone, welcome back to another episode of AI Proem Differentiated Understanding. This is your host, Grace Shao. Look where I am, the back seat of a car. Doesn’t look that exciting, does it? Let me flip this around. Look at that. There is no driver. I’m in the back seat of a Pony.ai robotaxi. Today joining me is James Peng, co-founder and CEO of the leading robotaxi company. It’s expanded its footprint across the globe, in Asia, in Europe, in the Middle East. But obviously, today we’re in its leading home market, China and Shenzhen, where it has a fleet of a couple hundred vehicles deployed on the streets already. Hi James, thank you so much for sitting down with me. I’m really excited to be having this conversation with you. So you left Baidu in 2016 to found Pony.ai when many in Silicon Valley were saying self-driving cars are only three years away. But obviously that wasn’t the case.Grace Shao: So, what did you believe then that this consensus was getting wrong? Tell us about your journey from 2016 until now.James Peng: Yeah, sure. We were founded in 2016, about 10 years ago. But even at that time, I didn’t believe that autonomous driving can be solved in three to five years. Just from a technical point of view, because even back then, 10 years ago, even a demo for autonomous driving was already very hard. Later on, there’s complexity involved in the autonomous driving industry that involves regulation, user acceptance, the readiness of the ecosystem. So because of the sheer complexity, even then, my prediction was it’s going to take at least a decade for this to be a real application. It turned out to be that my prediction was about right. Now, 10 years down the road, we actually have fully driverless commercial applications in many cities. Of course, it’s just the beginning of the long journey for autonomous driving. But at least now we have real commercial applications.James Peng: So I think people, like any new industry, people were super optimistic for the short term, but they were underestimating the potential for the long term. So I think autonomous driving is definitely one of those industries.Grace Shao: What really drove you to want to actually work on this, work on this technology and the future mobility?James Peng: I think the motivation was twofold. One is that the potential, both commercially and also societal benefits for the autonomous driving is so huge. Think about like everyone needs to have some sort of mobility. Autonomous driving is much safer than a human driver. So it has huge societal benefit of saving people’s lives. So essentially, it’s just such a great industry to work on. Although back then, 10 years ago, it was very unclear when this can be done. The other reason is, of course, because the sheer technical challenge of autonomous driving involves because I was actually in my previous jobs. I worked on different areas, software, hardware, large scale distributing systems, AI and whatnot. But none of the things I worked on is as complex as autonomous driving, which is a field that involves hardware, software, hardware and software integration and Many other things. There’s AI, there’s real time system, there’s also large scale AI training and all that.James Peng: So just from a sheer technical point of view, it’s such an amazing and challenging thing to work on. So I think those two reasons propelled me to start the company.Grace Shao: There’s definitely a lot to unpack there. I think later on we can definitely double click on the hardware, software integration, as well as the safety concern there. You say that autonomous driving is much safer than humans. For sure, it’s safer than me driving. I know that. But some may argue otherwise. So let’s talk about that later. But first, I want to ask you about something that was quite interesting. During 2020-23, there was a bit of a public reckoning, I think, within the industry. A lot of peers folded during that time. People decided to pull out of this sector. Some people worried that autonomous driving would really become a reality. But you guys charged ahead and you really believed in your vision. Tell us about that period and how maybe that changed your vision or your growth mentality.James Peng: I think 2020-23 was a period of time where the autonomous driving industry has evolved for roughly 10 years. I think that was the time of reckoning. That’s the time where the haves and have-nots have really diverged. So I think that’s actually exactly the time. As a company, we have seen tremendous progress. At the end of 2022, beginning of 2023, that was the time we actually finally had the first fully driverless commercial applications operations on the road. So because we made such progress, both from a technical and also from a regulatory point of view, that, of course, we made the progress. We finally see the glimpse of hope. Then, of course, we charge ahead. I think a lot of the other companies who weren’t able to, either from a technical point of view, or from a pure capital-raising point of view, or from a regulatory approval point of view, that weren’t able to have fully driverless applications. Then they were faded away.James Peng: So it’s sort of like, well, everyone is in school. There’s no big difference. But after graduation, then there’s haves and have-nots. So I think that was the time of division.Grace Shao: Yeah. So speaking of milestones, I want to kind of go back into history a little bit. So in 2021, Pony.ai had the third highest number of miles driven behind Waymo, Cruise. In 2022, Pony.ai became the first autonomous driving company to get a taxi license in China. In 2023, Pony.ai was licensed to operate robotaxis in Guangzhou, etc. And expansion continued. So kind of following what you just said, there was good momentum behind you guys. Now, today marks Pony.ai’s 10th year officially. You kind of talked about how you guys have grown. But what was one or two of the biggest milestones that you’re really proud of looking back now and that you think have really set the tone for your company Now as you are really expanding globally?James Peng: Yeah, I think in my view, the biggest milestone, actually, I have already mentioned, is the end of 2022, beginning of 2023, where we were granted the fully driverless commercial license in both Beijing and Guangzhou. We start to have the operation to the general public. Actually, it was in mid-January in 2023 that I was the first road in our commercial robotaxis operations in Beijing. Surprisingly, it was exactly on that day, it was snowing in Beijing, and I was in the vehicle by myself. That was the moment where I actually saw our vehicles were able to drive by itself. Anyone besides me in the vehicle. Because of the snowing, it was also a very challenging scenario. We were actually not being suspended for operation. We continued to operate, and I was in there. That was the moment. Finally, it felt like a dream come true, right? Finally, it’s not just because our technology is ready.James Peng: Also, because we actually got the approval from the government to have the license to operate. So it’s like all the seven plus years of efforts finally pays off. To me, that was felt like, as Lyndon Johnson said, the small steps for a person, but a giant leap for the human race. Although I wouldn’t call it as big as the Apollo, but to me, it felt like it’s finally from zero to one. So I think that was a deciding moment or defining moment for Pony.ai.Grace Shao: That’s a personal Apollo moment. I love how you visualize it because I could just imagine how chaotic the roads were in Beijing. Also to be quite romantic when Beijing is snowing because it’s such a beautiful city. Okay, but let’s talk about what is a robotaxi and how the public actually even felt about it when it first rolled out. Before we started recording, Ivy was even telling me, I was like, hey, look, I get a bit scared when I see Waymos on the roads or Pony.ai vehicles when there’s no one Driving behind the wheel. Now, that’s because I’m not used to it. You said, oh, yeah, it’s okay. People get used to it eventually, right? But let’s look back at 2022 when it first was deployed to the public. What’s the public’s reaction?James Peng: I think because it was a gradual process in the operational domains, in the operational zone that we had. We used to have a safety operator behind the wheel, although the driver actually didn’t touch the wheel or push the pedal. But people gradually get used to it. Actually, at the very beginning, when we were just deployed in Guangzhou, in those days, if you look at the picture of our first and second generation of Autonomous driving vehicles, you still see those spinning LIDARs on the top, and they were very much visible. People were curious. But gradually, people are just getting, this is like business as usual. As a rider’s point of view, the experience of a robot taxi is exactly like a typical taxi. The only difference is there’s no driver inside the vehicle, right? So the way you get the vehicle, the way you get in and get out is exactly the same.James Peng: Also the other traffic participants, like the pedestrians and cyclists, they get used to it. So I think it just takes time. It’s just like the first cell phone comes out, the first real smartphone comes out. People were very curious. Now it’s just, nobody cares about it. So I think it just takes time.Grace Shao: It normalizes eventually, right? Absolutely. I think we’ve really had a few years of consumer education done by quite a few of the players, including yourselves. Okay, well, let’s talk about the technical side of things. For outsiders, people might not understand the nuances between L2 and L4. Increasingly, we’re getting closer to L5 supposedly, are we? So help us understand your thinking on there. How do you structure your own teams, your products, working on different technology? Who gets held accountable for the actions in an L2 vehicle versus an L4 vehicle? Then finally, are we getting a glimpse into the future of L5? Are we going to be able to complete the road anywhere we want with autonomous vehicles? It’s a big, broad question, but I’ll throw it to you.James Peng: Yeah, sure. So the definition of the level of automation for vehicles was actually defined about 20 years ago. So, of course, the industry evolved quite a bit. I don’t think that the levels from L0 to L5 might be the right way of defining what the level of automation is. So in my opinion, actually, there are two different products. One is what we call the driver assist systems, ADAS. The other is fully driverless. So in a broad sense, I think there are two categories. There are definitely two different products. The biggest difference is not on the technical side, but rather, as you mentioned, probably on the regulatory side, is who is first in line for the responsibility if there is ever an accident. I think for any ADAS system, any driver assistance system, it’s always the driver behind the wheel that’s responsible. Regardless if he or she is looking at the road or has their hands on the wheel.James Peng: Whereas for the fully driverless systems, it’s the system that’s first in line. Because of that requirement, right? Think about if there’s a driver behind the wheel, it sort of serves as a safety net. So the system does not need to be bulletproof. It’s probably, well, as long as it can handle 99%, the case is probably good enough. Whereas for fully driverless, it has to be dealing with all the edge cases, all the extreme cases, and have a fallback system. We can get into those details later. But essentially, in my opinion, there are two different products. Of course, for the driver assistance systems, there are different levels, right? You can be, say, only highway or there’s only keeping in lane. Or they will actually even be able to handle some of the automations in the urban environment. For the fully driverless, of course, as you mentioned, there’s L4, L5 in a traditional definition. L4 means in certain areas. It can be fully driverless. L5 is everywhere.James Peng: But I think it’s never a clear division. Essentially, you can think of it as how we drive, right? We start with the area, then we gradually improve. Eventually, it will be everywhere. So I think that will be a gradual process instead of a clear division.Grace Shao: So actually, I want to double click on what you just said. So then help me understand, what is the gap between L4 and L5? Right now, Pony.ai is at L4, right? They’re robotaxis. Is that correct? How am I understanding this?James Peng: No, I wouldn’t call them a gap. I think it’s a different product definition. Because they serve different purposes. I think most people view this as a process of evolution, right? From L0 to L2, L3, L4. But it’s actually a wrong way of looking at it. As I already mentioned, because the clear difference is that who’s first in line with responsibility. That’s decided by regulatory, actually. By product definition. By regulatory as well. So because of that, it’s essentially two different products. As the product is getting more and more mature, getting more powerful, in my personal view, the division of two different products is getting wider instead of narrower.Grace Shao: Okay, then I’ll push on this. Then what is the real bottleneck right now for companies like you to deploy at a faster scale? Or to go into more cities quicker?James Peng: I think that’s the reason that I wouldn’t say it’s one single blocker or one single bottleneck that prevented us to grow faster. I think it’s because the sheer complexity of the autonomous driving and what entails to ensure safety. There’s regulatory, there’s technical things. We also, because this is such a brand new system, that we need a manufacturing capacity. We need deployment. We need to get all the operational things ready, like all the garage space and whatnot. Also user acceptance, user education, as we already mentioned. I think all those take time.Grace Shao: I believe also we have different partnerships with different managers of your local fleets. That kind of know-how also takes time for them to understand, to transfer over, right? For them to manage robotaxi fleets versus human fleets.James Peng: Absolutely, absolutely. All those takes time.Grace Shao: So I want to bring it back to technical. You have said publicly that you use the least compute footprint to reach L4. I thought that was quite interesting. Help us understand how you achieve that. How the model on the vehicle versus the large model you train in the labs actually work together.James Peng: I think all the AI systems more or less take the same approach, is that you have data on the backend, on the data center side. You train a large model where you essentially try to get all the cases to be learned. In our case, we use the word model, where you can think of it as a simulated city, where we train the virtual driver and let us drive on all different kinds of roads and learn the driving ability. So that’s what’s condensed as a model from all the learning that we deploy on the vehicle. In the traditional AI sense, that’s called edge computing. You put it on the edge, put it on the devices, and put it on the car, where it’s a much smaller model. In a human sense, it’s like we learn everything. Then when we go to a test, we don’t need everything. We just need to be able to have the ability to handle the test.James Peng: So that’s typically the training, where the backend system needs a lot of computing, but on the actual usage side, you don’t need that much computing power. So when the car is running on its own, it’s actually only using the model on edge, essentially.Grace Shao: Absolutely. I see. Okay, so let’s talk about AI systems, because AI systems for language, images, code have improved dramatically. There’s also obviously a lot of hype right now around world models, but what you’ve been describing actually has been something that’s not been coined world models for a decade, over a decade. What has generative AI done for you guys? How has it changed how you view your own AI system? Do you think, I guess, the word world models do your system justice in that sense?James Peng: Yes, it’s a little bit different, and they’re also related. Again, use human as an analogy. It’s actually quite similar to how we think, right? Because think about the large language model, how we process image, how we process knowledge. It’s sort of related to our memory and our logical areas of the brain. Whereas when we drive, it’s not just the memory and our knowledge. It’s also how we react, how we action, and all that. So the example is, one is related to how we learn a new skill. That’s the language model. Whereas for driving, it’s like how we learn to ride a bike. It’s actually different types of brain, different types of skill sets. That’s why it’s different. It’s not the same AI, because that’s how humans deal with different skills for knowledge.Grace Shao: So Gen-AI has not affected you, but how do you view the idea of now calling, I guess, the physical AI world, world models? Because you guys have been doing this for more than a decade. That’s kind of my question, I guess.James Peng: Yeah, that’s why I’m trying to get to it. For language model, it’s related with knowledge, with language, with logic. That means you need to be a very large model. Because think about it, if you don’t know a historical event, there’s no way you know it. So you have to have all the knowledge of a human ever created in your model for it to be powerful. So that’s why a large language model requires a lot of computing power and memory and everything. Whereas for driving, that’s how we learn riding a bike. We don’t need to have a PhD degree to learn how to ride a bike. But rather, it requires a lot of practice and training. That’s what world model is related to, or is assembled like. It essentially is a model where the virtual driver can start learning by itself, to learn how to interact with other cars, cyclists, pedestrians, and then learn the driving skill out of that.James Peng: So it’s a bit related with the large language model, but it’s quite different. Because it’s related with action, related with manipulation, related with collision avoidance. So that’s the key for the world model.Grace Shao: So tell us about how you simulate these systems. How do you leverage simulation systems for these edge cases?James Peng: So essentially, that’s how we learn how to drive. There are several key factors for the world model. One is it needs to be very real. So we call the fidelity. It needs to have high fidelity, means it resembles the real world. Second is everything that moves in the world model, meaning cars, pedestrians, needs to be smart. Meaning that because the thing that’s driving is the interactive process. Our action, because we constantly make decisions in there, our action will affect everybody else around us. So they need to react accordingly. So it’s interactive. It’s more like interactive gaming, where we react with everything else. So that’s the second challenge is all the interaction needs to be smart, needs to be intelligent. The third challenge is how do we evaluate what is a good driving? You can interact and everything. You avoid collision. Is that a great driving? No. Not enough, right? Because there’s a passenger inside. Comfort is important. Efficiency is important.James Peng: From A to B, we want to use the minimum amount of time. So essentially, it’s a multi-metric evaluation system in there to see what is a good driving. So there are three key challenges for the world model. We certainly, all our effort developing the world model related with that three areas.Grace Shao: But human drivers can be so emotional, right? Either you can be scared or you can be rage driving. Or we can be communicating sometimes without obvious signs, right? We’re looking at each other. We communicate with eye contact, hand gestures. How do you train your fleets to understand human behavior right now? Because obviously, human drivers are still the majority of drivers on the road today. In your case, I actually think you’re right. At one point, maybe removing all the human drivers will make it even safer, right? Especially removing drivers like myself, if I say it again. But yeah, how do you actually help these cars understand all these non-obvious signals? Not someone quite directly clashing onto you. Someone forgetting to turn on the turn sign. Someone stop sign looking at you, waving to go.James Peng: Like all the nuances. Absolutely. See, that’s why the first thing is how we become a better driver. Essentially, it’s a continuous learning process. The first thing, that’s how we learn how to drive, right? The first thing is you avoid collision. You were a cautious driver. Then gradually, you start learning a bit of everything else. All the signs, all the nonverbal cues, and the hand gestures. So that’s exactly the case for us as well. The earlier model of our system is just driving and try to avoid collision. Then gradually, we put a lot of new things, new recognitions, new perception models into our system where we start recognizing, for example, the hand gestures, especially all the policemen, all the typical police gestures, stop, Go, and all those things. Then we start recognizing, for example, potholes on the road, small obstacles on the road. So it’s sort of how we learn. We start getting all the big pictures first.James Peng: Then we start learning all the nitty-gritty details down the road and put them to enhance our system. Regarding the second point, you’ll see, when all the cars are autonomous driving by themselves, it will be easier to drive. Yes or no? Because the thing that the big, especially in China, in the roads in China, the biggest challenge is not the other vehicles. In a lot of cases, it’s cyclists and pedestrians. While we can’t make them to be autonomous driving, so I think having the ability to recognize pedestrians, recognize the intention, their sign, and give You a specific example on a crosswalk, the way pedestrians look at and how they pay attention. For example, if they want to directly cross, they typically look straight. But if they were looking back, that means they will more likely not to cross the crosswalk. So we actually take those cues to decide whether we let them cross or proceed straight ahead.James Peng: So a lot of those details need to be put into the system to make it safer and at the same time efficient.Grace Shao: Is the judgment made on the spot using the cameras?James Peng: Yes, using all the sensors. Cameras and LiDAR provide the sensor input.Grace Shao: We take it all and then we make the comprehensive decision based on the input. Definitely China has more complex and less predictable driving conditions, especially given the number of pedestrians, cyclists, motorcycles we just talked about. So if you can drive safely there, I bet you can drive safely anywhere. But jokes aside, it’s really interesting because, we talk about as your fleet grows, you accumulate more and more world data, real-world data. Is that kind of data eventually becoming an advantage and a serious edge for incumbent fleets and a structural barrier that makes it very hard for new entrants to compete then?James Peng: Data is important, but data is not everything. So how we understand that is this. Probably give you an example. Think about how we learn. Let’s say we learn math, right? You can think of the data is like the practice sets that we have. Of course, you need to do enough of practice to be a good knowledge about the subject. But doesn’t mean that you have the problem sets of the whole world that you become math experts. So that’s exactly the same case. We need enough of data sets in order to know what the real world driving condition looks like. But we don’t need everything because once we know enough, we can always generate enough knowledge about the driving. So in a way, I think the driving data is important, but it’s not everything. So that’s exactly how you view this.Grace Shao: So as we speak of this, how do you view the whole landscape right now? Who would you say are your biggest competitors globally? How do you view the different markets playing out?James Peng: Yeah, that’s a very complex problem. I think a question to answer because I think that I think the first and foremost, I think maybe I got some premises on this. First, the whole mobility industry, especially related with autonomous driving, is very large. They certainly have enough room for several players. Second, it’s still at a fairly early stage for the fully driverless. I don’t think the landscape is already divided in the set. So giving that two promises, I think currently, when I look at the players, I have to judge their current deployment. Although everybody can say, oh, they will have, they will, they will have thousands, hundreds of thousands of vehicles on the road. Actually, giving the current situation, I use the metric as having fully driverless commercial operation as a baseline. Giving that as a factor, I think in the US, Waymo is definitely leading the way. Because Waymo already have 4,000 or 5,000 vehicles on the road, 4,000 plus.James Peng: Then, of course, there are some other players trying to play a catch up. Zoox, Cruise, maybe Tesla as well. So there’s, of course, some. So I would say in the US, Waymo is leading the way. There’s three to five players trying to play a catch up. In a global sense, I think from a technical point of view, China’s player is certainly on par with the US players. But from the total cost or the economical sense of a vehicle, for example, our vehicle is four or five times cheaper than Waymo’s vehicle. So in the global markets, such as Europe, such as Middle East, I think we will play a huge edge compared with the US players. Certainly, the whole landscape is still evolving.Grace Shao: But especially in the global markets, you’ll see we will definitely not play a catch up, but taking a leading position. You’ve been an advocate for hardware optimization, software optimization, battery solution optimization. Is that the strategy behind being able to have a vehicle that’s four to five times cheaper than Waymo’s? Or where’s the edge? Or how are you building these comparable vehicles at a relatively cheaper cost?James Peng: Yeah, I think as you mentioned, you definitely mentioned the most important factor to have the much cheaper price on the vehicles, which is optimization on Hardware, software, and everything else. I think another reason, of course, is because the whole ecosystem related with autonomous driving in China is relatively mature, and the scale is larger. So that price is cheaper. For example, the vehicle itself, the sensors, they’re relatively cheaper in China than anywhere else. Because of the ecosystem, because of the scale. So that plays an important role as well. That touches on something. A lot of physical AI, a lot of robotics companies are also now leaning into the Chinese supply chain. A lot of your peers, actually, autonomous driving, or even the EV players are now looking to expand into physical AI, whether that’s robots, humanoids, Quadrupeds, whatnot.Grace Shao: So you’ve stayed really focused. You’ve not launched any robots out there or anything. What’s your thinking behind this?James Peng: Yeah, absolutely true. I think autonomous driving definitely is probably the first large application of physical AI. All the others, humanoids, robots, and everything else, probably will have real applications down the road. For us, we view the autonomous driving as our brand and partner. Of course, as I mentioned, this is still early stage. I think we still have a lot of mileage to go. For all the other physical AI applications, we don’t have any specific plans yet. But I think they’re definitely interrelated. We may enter them down the road, depending on whether we need it or not. Because my judgment is that for the physical AI, it probably will follow the similar trend as autonomous driving. It might take another decade for it to mature. I think for us, it’s more like whether we have real applications for it. Give you a specific example.James Peng: Even for our fleet, once we go to hundreds of thousands, millions of vehicles, how we maintain those vehicles, how we do the charging, cleaning, servicing, They may use robotic applications. So I guess my view is that we will not probably do robotic actions just for the sake of doing it. But we may do the related applications when we see the real applications.Grace Shao: So it’s fair to say you’re cautiously optimistic that there is a potential use case further down. But it’s nowhere close to where it’s been hyped in the three to five years kind of use case.James Peng: Yeah, I think it’s the same thing as autonomous driving 10 years ago.Grace Shao: All right, well, let’s talk about your international footprint. You mentioned earlier, you have a global strategy. You’re in Europe and Luxembourg was your first launch, right? You’re in Southeast Asia, parts of East Asia, you’re in the Middle East growing really fast over there. Tell us about how you think of your next steps in your global expansion.James Peng: Yeah, I think the mobility demand everywhere is the same, right? There’s a strong demand across the globe. But we have to focus on the most important markets first. I think eventually we’ll go everywhere, because that’s our motto is we have autonomous mobility everywhere. That’s our ambition when we started 10 years ago. But our first launch, we have several criteria. One is related with regulatory, right? It needs to have relatively accommodating regulatory environment. Second is it needs to be a relatively mature mobility market. In a more obvious sense is that the local taxi fares needs to be relatively high, because I think that our pricing anchor point is always a human driving Taxi. So that price needs to be relatively okay. The third criteria are that we need a good local player to partner with, because a lot of other things like regulatory, like the back end services needs to be Handled by the local partners.James Peng: So judging from that three categories, I think Europe, Middle East, Southeast Asia, Japan, South Korea, Australia maybe. Those will be probably the potential markets for the initial launch. Of course, those are already big enough of number of countries. So we’ll pick and choose some to start with.Grace Shao: How do I understand your partnership models? Because I believe you’ve quite a few different kind of models depending on the location, the regulatory environment, potential partnerships, know-how, etc. Tell us about that.James Peng: Maybe I’ll take one step back first. Think about what is a robotaxi industry. The type of players, I’ll divide them into four categories. One category is for the user acquisition. Those are ride-hailing applications. Those are the Ubers and the Lyfts and the DDs alike. The second is a vehicle. You need a car, how you manufacture a car. The third is a driver. The fourth is all the back end services, cleaning, charging, servicing, insurance, and everything else. For us, our main job is creating a virtual driver, is making a really safe, efficient driver. So that’s definitely what we do. All the three other categories, we might have partners, we might do ourselves. So that, depending on the market, depending on what’s the strong local players, we might pick and choose players who’s handling one or two or three of the Other things. For example, we work with the ride-hailing platforms for the user acquisition.James Peng: We work with some of the back end services who’s providing the parking space, who’s cleaning, charging our cars. We also have OEM partners that work on the cars. So that’s how we view the partnerships landscape.Grace Shao: So after you deploy, say you send these out to Australia, what happens walk us through that. Because once these cars actually get off the boat and ships and they land in Australia, are they your responsibility or your partner’s responsibility? Do you send an engineer? Do you send your own management? Or do you transfer that know-how and maintenance know-how to the local partners to handle?James Peng: Great question. That really depends on the different partnerships and different regulatory environment. In some markets, it’s the local player who’s first in line with managing the fleet. That means in those cases, we manufacture the cars with OEM. But then once we ship the vehicles to the local country, Australia, giving you an example, or Singapore, let’s say, then we actually, in those cases, we sell The vehicle to the local partner. It’s like selling hardware. It’s like selling hardware. But we will, of course, have engineers handling the driving because we are in charge of the driving. So all the driving related work will be done by us. But then the user acquisition, the cleaning, servicing, charging will be done by the local partner. So those are one case. But in some markets, we actually ship the vehicle and we apply licenses by ourselves. The vehicle is still on our own book. But those are rare cases.James Peng: We actually, our preferred model is to have the local partner that handles most of the logistics and we will be the tech providers. We’ll essentially have the virtual drivers handling the driving and everything else is done by the local partner.Grace Shao: I see. Would you ever view OEMs as competitors in any way? Because right now you’re partnering with them. You’re giving them the software enablement, right? Would they produce their own robotaxis?James Peng: I think in most cases, in my view, that they probably will be partners instead of competitors. It’s very different because they were mostly on the hardware business. Very few of them will be in the robotaxis business because they’re quite different.Grace Shao: I see. I see. Something a bit niche is, I know you run robotaxis as well as trucks. Walk us through how you think about that. Why do you guys also have a truck business? What kind of scenarios are they already being deployed in? I believe they’re the heavy trucks and then the light trucks. How do I understand this?James Peng: Yes. As I already mentioned, think about our business is that all our technology is that we are creating a safe virtual driver. Virtual driver is our core. As a driver, you should be able to drive all different types of vehicles. The two biggest applications for driver is one is for the transportation of human beings and the other is for goods. That’s related with robotaxis and then for all the trucks. Within the logistic industry, there are actually three categories. One is for the long haul, which is typically done by the heavy trucks, the 18 wheelers and whatnot. Then there’s also in-city network, which is the light duty truck. Then there were also the last mile, typically is handled by much smaller vehicles. Our main focus, of course, is on the long haul and the intra-city transportation. On the last mile, we are the providers of the domain controllers, but that’s not the areas we’re working on. So think about we’re creating driver.James Peng: Driver should be drive different types of vehicles. That’s how we view the trucks versus the robotaxis.Grace Shao: Usually, I would assume these are like ports, airports, maybe?James Peng: They will eventually be everywhere as well. We start with ports. We start with some dedicated routes, for example, like a minefield to the local distribution center, those 30-50 mile routes. The reason we start with those applications is because typically it’s mostly because of regulatory reasons. Because the ports and the dedicated routes and those are typically a semi -private road. It’s much easier to get the regulatory approval. Of course, we are working on long haul trucks as well. We already actually have a fleet of heavy-duty trucks doing the real goods transfers on highways, but still with a safety driver, of course. Eventually, we’ll be fully driverless as well.Grace Shao: You’ve said you have a target of running fleets commercially across more than 20 cities by the end of this year. What do you know today that you could not have learned without actually operating at scale already on the streets? What makes you have the confidence to do that now, I think, compared to maybe a few years ago?James Peng: Again, I think for robotaxis commercial business to be a reality, there are three important factors. One is technology. Second is regulatory approval. The third is user acceptance. I think within the last three to four years, we have gained a lot of experience on all three categories. The reason we were confident to deploy in 20 cities is because clear vision on the regulatory approval. There’s a lot of cities globally, both in China and in some global cities, they actually start coming out with regulations for supporting fully driverless commercial applications. Also we have planners. Planners want them. So I think all the important factors are falling into place. That gives us confidence.Grace Shao: I’m going to play devil’s advocate a little bit here. With the rise of AI right now, there’s a bit of a fear of replacement of people’s jobs. The rise of autonomous driving obviously lead to job loss in people who are currently drivers. How do you view that? Because just now we talked about robotaxi drivers. We talked about people driving heavy-duty trucks that could potentially be replaced. Frankly, I’m in a camp that people could be maybe freed up to do more things that they can do otherwise. People will find alternative careers. But are regulators becoming more cautious. How do you feel about the current public pushback a little bit on AI, autonomous driving, autonomous everything at the moment?James Peng: Yeah. Actually, driving is a hard job. Driving is a lot of cases in a stop vehicle for 10, 12 hours a day. It’s a really tough job. The thing that because autonomous driving itself is a highly regulated industry, the pace of our roll up is determined by the number of licenses. The thing about also a lot of the drivers were not young. The young generation, younger generations actually don’t want to be drivers. So I think, especially a lot of the global markets, we actually come in to fill the gap for the labor shortage for the driver. We’ll not change the human driving vehicles overnight. It will be a gradual process. So that’s sort of the development of the cities and the human society. It takes time. It becomes gradually a norm. Then, as you just mentioned, then the drivers can find other jobs.James Peng: Even we actually absorb a lot of jobs, for example, for the remote assistance, maintenance, which are much safer and much less strenuous job conditions. So I think society as a whole has always a way to absorb jobs. To adopt, adapt, and then evolve.Grace Shao: The current pay for a lot of times for these heavy truckload drivers are like 200 to 300k USD. They’re considered very high-earning jobs. But at the same time, people forget they’re extremely dangerous. There’s life lost constantly on the roads. So I can see that could be very valuable if people can actually replace those routes with robo-drivers.James Peng: It’s not just replacing. Look at the truckers. Their average age is 45 plus. In North America right now? In North America. In China, they’re 40 plus as well. So a lot of younger generations, they don’t want that type of jobs. We’re coming not only to replace, but actually to fill the void for that job shortage.Grace Shao: All right. So I think I want to wrap up our conversation soon about this. Is there anything I’m really missing, you think, about robotaxis and your business at this point?James Peng: I think we’ve probably covered a lot of topics.Grace Shao: Oh, I had one question. Another one about your business before we go into your personal thing. You mentioned Croatia just now when we were talking offline. I thought that was so fascinating. In my mind, I thought these robotaxis were being deployed mostly in futuristic cities like Silicon Valley and SF, out here in Shenzhen where we’re here today. But Croatia, help us understand the need for robotaxis in these countries where a lot of the roads are aged, are not really made for cars to start with, Are not easy to drive in, actually, even for humans. Then how does that make sense even for your economics, actually?James Peng: Of course, there were some challenges. From a technical point of view, two challenges initially. One is there’s a lot of roundabouts. Actually, there were not many roundabouts in China. So although a lot of other very complex situations like heavy storms and whatnot, we were able to handle them really well. But roundabouts, we had some, but we haven’t trained that much. So we actually have to retrain a bit on the roundabouts. The second is the trams. There were just a lot of trams in the Zagreb. Their behavior of the trams is different from cars. So we need a little bit more training to get used to it. But it’s like how we drive. When we go to a new city, we might not drive as a perfect driver initially. But then we learn and adapt. Once we have a good learning system set up, then we can quickly learn. That’s exactly our experience in Zagreb, Croatia. Two things that we actually have to learn in Croatia.James Peng: One is the roundabouts. The other is trams. Because those are not something that typically you will see on the roads in China. So for those new situations, it’s like how we learn. How we learn driving. When we go to a new city, we probably know 95%, 98% of the situation. Some of the scenarios probably we didn’t encounter previously. Then we learn. We adapt. So that’s exactly the case for us in Croatia. After three to four months of learning and training and retraining, we actually were able to handle those cases like roundabouts and trams really well. Because there’s a lot of roundabouts in other cities in Europe. They actually have different rules for roundabouts. Some of the roundabouts, I think the cars outside roundabouts have right-of -way. Some of the vehicles inside the roundabouts have right-of-way. But we can adapt once we have the system set up.James Peng: So as I mentioned, the most important characteristic of our system is not how powerful it is, it’s how adaptive and how easy to learn on our system so that We were able to adapt.Grace Shao: Brilliant. So a lot of localization as well for your vehicles. I have two last questions. One is, what is something you think people still get wrong often about your sector, in this case, autonomous vehicles, autonomous mobility? The second question is a bit of a curveball. I’ll throw it to you first, you can think about it. What is one differentiated view you hold? Something that’s a bit against consensus, maybe.James Peng: Autonomous driving industry, I think people put too much focus on technology and probably underestimated the complexity of robotaxi as a business. Essentially, of course, technical is the most important. If you can’t drive safely, you’ll not have a business. But once you even have the most safest driving, you still have to, as a business, there’s a lot of other things involved. For example, how you deploy a fleet, how you make the pickup and drop off easy for the user, how you handle all the edge cases of the complaints of the Passengers, how you make the charging, servicing, cleaning efficient. For example, especially give you a specific example, the electricity fares during the day fluctuates. If you have the charging at the low fare, you can save a lot of cost. Then how you manage your fleet? Although you have the low fare for electricity, but the demand of the passengers is really high. How do you make a decision?James Peng: So essentially, it’s a lot more optimization involved than just the driving itself. I think a lot of people underestimate the complexity with the management of a fleet of autonomous driving vehicles. We actually, as a company, have put a lot of emphasis and take a lot of efforts in optimizing everything. So that’s why I think those will be a very strong competitive edge down the road.Grace Shao: Once you guys scale further, especially.James Peng: Exactly, absolutely. Very interesting.Grace Shao: The second one, I’ll put you on the spot again. What is one differentiative you hold?James Peng: I think I’ll take the one related to the answer of my first question. Is that, again, people always put too much emphasis or give too much credit on zero to one and think about less for one to ten. Give a lot of examples, right? People always think an invention is so hard, but putting an invention to be a scaled application is equally hard or a lot harder. Because the scale involves cost optimization, involves user education, involves a regulatory approval, it involves making the things a lot easier to use. So many examples like this, right?Grace Shao: Definitely. Say a rocket is put in the sky. Oh, it’s so hard. But having the rockets to always be able to safely take off and recycle, that’s extremely hard.James Peng: So I think related with autonomous driving is we certainly crossed zero to one. I think we crossed one to five, maybe. But from five to ten, ten to a hundred, I think there will be still a lot of challenges ahead.Grace Shao: That’s very insightful. I agree with you. When we look at the internet era and a lot of players that still stand today versus who are the actual ones that created a lot of the internet use cases we know of today. Thank you so much, James. It was a pleasure and an honor to learn more about your business, yourself, the man behind the company that is changing the future of autonomous mobility. Thank you again.James Peng: Thank you for having me.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  From Beauty Apps to AI Agents: Meitu’s CFO Gary Ngan on the Future of Visual AI 14.07.2026 50λ

  In this episode, I spoke with Gary Ngan, CFO of Meitu, about how the company is evolving from its roots in consumer photo editing into a broader AI-native visual creation platform across photo, video, design, and agents. For many investors, Meitu is still associated with beauty editing and selfie apps, but Gary frames the company today as an AI application company serving both leisure use cases and productivity workflows.We spent a lot of time on Meitu’s business edge: why visual AI is not just a foundation-model race, and why aesthetic judgment, controllability, and vertical context matter. Gary argues that visual creation is highly subjective. The same prompt can mean very different things across countries, cultures, product categories, and commercial goals. That is why Meitu is building verticalized products such as Picchi, DesignKit, Kaipai, Vmake, and RoboNeo, instead of relying only on one general-purpose AI model.We also discussed the business model. Consumer subscriptions have become Meitu’s main revenue engine, while advertising is no longer the strategic growth driver it once was. Gary explained the shift in Chinese consumer willingness to pay for apps, the higher ARPU potential in overseas markets, and how new AI-native products like Picchi could introduce additional monetization through personalized models and AI credits. He also addressed AI compute cost, why more than 90% of Meitu’s AI outputs come from its own models, and why the company sees AI as a TAM-expanding opportunity rather than simply a margin risk.Finally, we covered competition and globalization. Gary explained how Meitu thinks about competing with ByteDance, Kuaishou, Canva, Adobe, Shopify, Alibaba, and other AI-native visual tools, and why Meitu’s approach is more vertical-driven than general design-platform driven. Lastly, we touched localization, from different beauty preferences across markets to why true globalization requires understanding culture at a much deeper level than translation or marketing campaigns. CHECK OUT THIS CONVERSATION. Gary’s so cool.To find the previous episodes of Differentiated Understanding, see here.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.Season two will host a series of guests from analysts, VC investors, builders, researchers, founders, and product managers. For more information on the podcast series, see here.Chapters:00:00 What is Meitu today? Mapping Meitu’s product portfolio04:14 Why vertical focus still matters in the age of AI agents06:36 Aesthetic standards, subjective prompts, and visual AI nuance11:31 How AI changes art and creative expression15:02 MeituHub and MiracleVision as visual AI infrastructure17:01 Why Meitu needs its own models20:55 How Meitu chooses models and the role of designers25:09 Meitu’s AI legacy and generative AI strategy28:27 AI compute cost, ROI, and gross margin38:22 Subscription growth and advertising dependence40:47 Partnerships with consumer chatbots and platforms43:27 Competition with ByteDance, Kuaishou, Canva, Adobe, and others49:49 Deepfakes, misuse, and AI safety safeguards52:42 Globalization, localization, and cultural differences59:16 The biggest investor misconception about MeituTranscript (AI-generated, for reference only)Grace Shao:Gary, thank you so much for joining us today.Gary Ngan:Hi Grace. Good to be here.Grace Shao:I’m really excited to have this conversation. To start, tell us what Meitu is up to these days. For a lot of investors and users, when they think of Meitu, they still think of the selfie and beauty-editing app. How would you define Meitu today? Is it still simply a consumer AI company, or is it much more than that now?Gary Ngan:Meitu is no longer just a selfie or beauty-editing company. I would define Meitu today as an AI application company specializing in photo, video, and design.We focus on very high-value verticals where we can leverage AI to deliver high-quality results to users. We often refer to these users as prosumers: people who have strong design needs, but no prior formal design training.So that is how I would define Meitu today.Grace Shao:That makes sense. Tell us about the products, because you have quite an array of them. Some are more consumer-facing, some are more prosumer-facing, and some may even be a bit more enterprise-facing. There is Meitu, BeautyCam, Wink, Picchi, DesignKit, Kaipai, Vmake, RoboNeo. Help us map out the ecosystem.Gary Ngan:We think about Meitu’s product portfolio in two main categories: applications for leisure and applications for productivity.Applications for leisure include the Meitu app, BeautyCam, Wink, and Picchi. They serve use cases such as photo-taking, photo editing, and video editing, usually for sharing on social media.The Meitu app and BeautyCam are our core consumer applications. Wink extends our capability from photo to video editing. Picchi is our latest portrait-retouching agent, focused on personalizing editing styles.The second bucket is applications for productivity, which includes DesignKit, Kaipai, Vmake, and RoboNeo. These products serve professional and commercial content creation needs.DesignKit focuses on e-commerce product-listing design. It helps merchants and creators produce product images, model images, and marketing materials much more efficiently.Kaipai and Vmake focus on talking-video and marketing-video creation. Kaipai is more focused on the domestic Chinese market in verticals such as insurance and real estate, while Vmake is seeing strong traction in the U.S. fitness and wellness market.To give you a sense, as of May this year, Kaipai had about three million monthly active creators, and cumulative content creation exceeded 400 million pieces. For Vmake, ARR in the first quarter of 2026 was about US$4 million.Then there is RoboNeo, our AI-native agent product launched in July 2025. It is currently targeting the AI short-drama vertical. Its agent workflows can support scriptwriting, characters, storyboards, visual generation, and asset management.So that gives you a rough idea of the different vertical products. But the ecosystem logic is very important, because many new products come from user insights we observe in existing products.For example, DesignKit came from the poster-design function within the Meitu app. Kaipai came from the AI teleprompter feature in BeautyCam. Picchi came from new user behaviors we observed in the Meitu app.So our portfolio is not a random collection of apps. It is a structured expansion from consumer imaging into AI-native workflows across photo, video, and design.Grace Shao:That makes a lot of sense. But in the age of AI agents, would it make sense for Meitu to consolidate a lot of these apps? Or do you still think it is better to keep them separate for different types of users and workflows?Gary Ngan:In the age of agents, we still believe we should focus on high-value verticals, because different verticals have many differences.First of all, aesthetic standards are very different. I’ll give you an example. The phrase “handsome guy” would be interpreted very differently in an application serving the U.S. market versus an Asian market.Even within the Asian market, if you are addressing e-commerce merchants selling gym products versus formal apparel, the word “handsome” will also be interpreted very differently across those verticals.So being able to separate these different verticals gives you a very good head start in focusing on the aesthetic standards that each vertical needs.Also, users in different verticals have very different behaviors and workflows. It is very important to build those workflows and that know-how into each vertical in order to create the right products.With agents, you can cover a slightly bigger boundary. But I still think you want to focus on different verticals to maximize the output for the user, and also make it more efficient and easier to market within each vertical.Grace Shao:That is really interesting. You touched on something that a lot of people discuss when they think about visual AI, which is how to ensure consistency and accuracy when translating language into visuals, especially when text can be in different languages and words can be subjective.As you said, if you say “handsome” and I say “handsome,” that could mean very different things in our heads. How do you ensure that identity, description, and nuance are not lost? You mentioned vertical focus, but what is the technical side of that?Gary Ngan:Instead of calling it fragmentation, I would say vertical focus is very important. That sets the tone.Behind that, we also have a large team of designers who control different points in the model fine-tuning process. They help set the right direction for the aesthetic standards within each vertical. That is something differentiated in our product offerings.Then, if you move one step forward, the data flywheel is also very important. Users within a vertical give us data through their behavior: which photos they use, which photos they edit, which ones they throw away. That is very important for us to improve image creation.We are also in the camp that believes controllability in visual applications should not just come from AI. You still need manual touch-ups at the end for users to make last-mile improvements, because aesthetic judgment is very subjective.Even if an AI model works with you every day, you will always have subjective comments and small edits you want to make.One other interesting point is that when we talk about aesthetic standards, in the case of leisure products, the face is usually yours. So you have a strong say and a strong sense of what is good for you.The way we learn that is by studying the trend in your geographic location, giving recommendations, letting you try them, and then as you use the application, you tell us what is most suitable for you.Picchi is a newly launched app where you can upload three to five sets of original photos and edited photos that you have done yourself. We are then able to learn that pattern and create a specialized model for you. The next time you want to edit a photo, you can call up the model that you trained yourself and apply your own aesthetic standard to your photos.On the productivity side, however, aesthetic is not the ultimate holy grail of an image. It is very important, but whether that photo or video is effective in driving conversion, likes, or comments is also very important.When we deliver images and videos to users, we take into account key data from that particular vertical and the metrics that matter for results. It is not just whether someone is subjectively handsome. It is whether this person, image, or video can help sell your product.So the two camps are quite different.Grace Shao:That is really interesting. From a pure consumer point of view, you pointed out an important nuance. If I upload my own face to a Meitu product, it might give me very smooth, pale skin and a more angular jawline or chin. But if I use an American fine-tuned product, it might give me more contouring. It is a very different aesthetic.But to your point, if you are a prosumer, a content creator, influencer, or e-commerce seller, then it is not only about whether the image looks good. It is about whether it drives sales.I want to ask something slightly more philosophical before getting into the businesses. Technology often changes art. Photography changed painting. Software like Final Cut Pro and Photoshop changed photography and video. How is AI now changing how visual artists approach their vision and craft?I have also spoken to companies like Kuaishou, where they have Kling and are partnering with AI-native film studios. These people are creatives, but they do not view AI as disrupting their work. They use AI as a tool to create their work. What do you think about this at a high level?Gary Ngan:If you look at our core value proposition, our mission statement is uniting art and technology.One step down from that, we are trying to democratize design, art, and creative expression.What AI changes is that it enables people who have creative ideas, but not the actual training or skills, to express those ideas.A lot of the time, we have creative ideas that we want to express, but our motor skills are not refined, or we do not know how to put colors together. AI can help us deliver those ideas.That is fundamentally changing the artistic landscape to a certain extent.Other companies may say that existing professional filmmakers and designers can use AI to make things more efficient or create things in a different way. That is great. But I think the bigger impact on the world is enabling many people who previously could not create anything. They had ideas, but could not express them. Now they are able to express them.That is what is fundamentally changing the industry.Grace Shao:We have talked about how you have many different products, and you explained that they are targeted at different verticals. How should we understand MeituHub and MiracleVision? Are they the operating system underneath everything?Gary Ngan:MiracleVision and MeituHub are the visual, image, and video infrastructure that we have. Our applications are built on top of these things, so they go hand in hand.We are still fundamentally an AI application company, but we also need visual infrastructure.To give you an example, over 90% of our AI outputs come from our own models. There are many situations where we think the models we fine-tune ourselves perform better than third-party models. There are things that other people do not necessarily focus on, so we have to invest in R&D and create that infrastructure ourselves.MeituHub is also a way for us to export that technology. People can use our APIs and skills to build their own applications or integrate them into their own systems. That also reinforces our vision of democratizing design.Grace Shao:That is a perfect segue to my next question. I understand your team fine-tunes your own models, but you also build on various open-source models. Why does Meitu need its own model?Traditional application companies often did not need to own the foundation layer. So why does Meitu need that? And more broadly, why are so many Chinese consumer internet companies pushing out models? You see even companies in food delivery, ride-hailing, and other consumer internet sectors releasing models. Is this a cultural push, or something else?First, how does Meitu think about it at the company level? And if you can comment, how do you view this competition across the China ecosystem?Gary Ngan:It is harder to comment on the overall market, because what we do, visual image and video models, is quite different from language models. So I will focus on why we do our own models.Our belief is that one general model will have difficulty performing well across all verticals, because context is so important.If we do not have our own models, then aesthetic standards will be set by third parties. When a third party creates a model, they have their own idea of what aesthetic standards should be. They have their own idea of what should be generally good given a certain prompt word.But that may not be applicable to the verticals we are working on. That is why we need our own models to serve those purposes.At the same time, we integrate third-party models because even within a vertical, there are corner cases or edge cases that our core model may not be optimized for. In those situations, we call on third-party APIs to serve users.As an AI application company, the only point of optimization is user satisfaction. We use a combination of our own models and third-party models to serve that purpose.Sometimes we see more and more users calling third-party APIs for similar prompts or similar creation scenarios. Then we will augment our models to cover those scenarios as well.Our model is continuously growing, but we make it very vertical-driven. We have told the market that we are not in the business of creating a general-purpose model. We are creating vertical models. But that does not mean we are giving up model training altogether.Grace Shao:So there is a lot of industry know-how in each vertical that you have.When it comes to which foundation model you choose for each task, how do you make that decision? I spoke to one of your colleagues at SuperAI, Rocky, your VP of R&D. We discussed the fact that you use a series of open-source models and also work with different model providers. What is the main factor in choosing which model to build on for which vertical? How do you delegate tasks across models?Gary Ngan:At a high level, there are two main forces behind that.One is user behavior. If a user uses Model A to create a certain task, and many users do not press save or do not continue working on it, then we probably need to serve that task with a different model. It is a data flywheel type of operation.The other factor is our large team of designers, who are very involved in training these models. Designers help set the standard for what the right model should be for a particular task.This is a very important differentiation for our company versus most technology companies.I am not sure if you are aware, but our founder and CEO was an art student by training. In his day, he was the top student in the Tsinghua Arts Academy entrance exam for oil painting.In his mind, aesthetic standards are always very important. Because of that, designers in our company have a very strong say in every product and every feature we launch.Over more than a decade of designer training, the rest of the company has also developed stronger aesthetic standards. Product managers and R&D engineers also have quite high aesthetic standards now.Our company is organized toward delivering the best aesthetic standards for users. That is very differentiated from most tech companies.Most model companies may think: We solved this problem, the photo is done, the video is generated, the main character is stable throughout three minutes, so the mission is accomplished.But for our designers, apart from the stability of the main character, they also look at whether the lighting is realistic, whether the color fits that vertical, and whether anything feels wrong from an artistic point of view.Those are the things we really focus on when fine-tuning. That is something we are very proud of, and I think it is a major differentiator.Grace Shao:Even as a consumer user, I can say your products have that extra last-mile touch-up tool that others often do not offer. It is meticulous and accurate. You can zoom into pores or details in the background. It is interesting to hear about your founder’s background and that artistic legacy, because that culture really shines through the products.Speaking of legacy, I want to understand Meitu’s AI legacy and strategy. You have been working in image and video for over a decade, so you obviously have a vast database and deep know-how in visuals. How does that industry expertise translate in the age of generative AI and in the future agentic world?Gary Ngan:Generative AI has changed the speed, scope, and value of what we can deliver.First, speed. New AI capabilities can now be translated into user-facing features much faster, helping us launch popular effects globally and drive overseas growth.Second, user experience. Generative AI enables effects that traditional computer vision technology could not fully achieve.For example, facial and body retouching is no longer just manual adjustment. AI can reconstruct details, lighting, and texture in a much more natural way.Third, target addressable market expansion. AI helps us broaden into productivity workflows like DesignKit and Kaipai, which were things we traditionally could not do.Overall, AI is very empowering in helping us get to where we want to go.Before AI, all we could deliver was better tools. But in order to use those tools, you still needed pretty good aesthetic standards or some understanding of the basics. Otherwise, giving you those tools did not really help much.With AI, you still need maybe 10% or 20% of that understanding, but the requirement is reduced massively. AI can give you many choices to choose from, and then you can start building from there.That helps us move from leisure applications to productivity applications. That is really what the strategy is about today.Grace Shao:AI can act like a guide or mentor if you are new to a certain craft or sector.Let me ask the spicy question. AI compute cost is obviously extremely high. Image and visual generation are expensive. How does the economics work right now? Does AI compute affect your gross margins, or are you seeing ROI already?Gary Ngan:As I said, currently over 90% of our generative outputs come from our own models. As long as we are using our own models, the cost is very manageable. Our gross margin is still over 70%.Also, when you are editing your own face or editing a product photo, these things are not purely AI-generated. You may want AI to edit a little bit, remove someone from the background, or create a new background for a product, but the entire photo is not purely AI-generated.It is true that AI inference has a cost, but it is not as if every photo now incurs a lot of cost. We need to make that distinction first.As we move into new verticals, like music videos and AI short dramas with RoboNeo, those are more experimental. We are using more third-party models, so margins on those new applications will be much lower than something like Meitu Xiuxiu.But as we continue to progress, we will develop our own models to replace some of the third-party costs. Over the longer term, we also believe API costs will come down.So we do not see this as a threat. In fact, the integration of AI has expanded the addressable market so much that it is a much bigger opportunity than threat.Grace Shao:I appreciate that nuance. You are explaining that the first type of usage does not use as much AI or token cost as people might expect from the headlines. The second part may be more expensive, but we are still in very early stages.Let’s take a step back. For some of our American or Western audience, they may not be as familiar with Meitu. How do you fundamentally make money?In your public disclosures, consumer paid subscribers grew more than 30% year-on-year. What is driving that growth? Is it that the AI features are much better now? Is it global expansion? Help us understand the business model and what is driving growth.Gary Ngan:Our main revenue source is subscriptions, mostly on the leisure side.That is our second growth curve. The first one was advertising, but that business has matured.The second growth curve, which is still growing quickly, is subscription on the leisure side. The main driver has several parts.The first is China user behavior. Paying for apps really started after COVID. Before COVID, virtually all applications were free. They competed through free usage, advertising, or redirecting traffic to other applications to generate money.After COVID, many user-facing applications realized advertising was under pressure, and they wanted new revenue sources. Without colluding, many of them started charging users. That kickstarted the user subscription process.What is less understood is that users then began realizing that applications have to be paid for. As time goes by, the behavior of paying for applications grew on them.Now there is much less of the issue of, “This app has to be paid, so I am not using it.” That was a real mindset before. Now it is more like, “This app costs 15 RMB a month. Is it worth it?”That is what I would describe as the beta factor, meaning the overall market. Users are becoming more and more used to paying for mobile products.That is one reason we are confident that paying subscribers and the paying subscription rate of our leisure applications can continue to grow.To give you a sense, we have done surveys. The global paying percentage for photo and video applications is about 20%. If you benchmark music and video apps globally versus Chinese equivalents, the Chinese equivalent is usually around half. For example, if Spotify is around 40%, the Chinese equivalent might be around 20%.So if global photo and video applications are at about 20%, China should at least achieve about 10%. Right now, we are around 5% to 6%. So there is still another 80% to 100% growth headroom there.The second growth potential is international expansion. In high-ARPU areas like the U.S., Europe, and East Asia, including Japan and Korea, the base ARPU is already much higher than China, anywhere from 100% to 200% higher. The paying percentage can also be much higher.To give you a sense, one of our applications called AirBrush has over 50% paying percentage in the U.S.As we launch stronger operations in these high-ARPU countries, we expect our blended paying percentage to grow further.One final point about monetization is that we are integrating more generative AI capabilities into these applications. For example, Picchi is an application for leisure, but it uses an agent for editing, and that has a completely new business model.On top of regular subscription, if you want to create your own model to apply your own editing skills, you have to pay for that model separately. That is another monetization test we are currently working on.Grace Shao:When I was reading your earnings reports, I was a bit surprised that your highest revenue generator is consumer subscription, because the default mindset is that people have very little willingness to pay.But as you said, whether it is the change in behavior in China, or people having more appetite for premium add-ons or AI-plus features, willingness to pay is changing.There is also the fact that advertising can be annoying to sit through. You do have a lot of advertising, I have to say. Spotify does too, and I think that drives people to pay to get rid of advertising.On that note, do you think you will gradually reduce your dependency on advertising? It is still your second-largest revenue model.Gary Ngan:We have not relied on advertising since 2022. At the corporate level, we made the point that we are no longer strategically trying to drive advertising.You have seen our advertising business grow at low single digits over the past few years. Advertising is not what we are fundamentally trying to drive.However, we are experimenting with advertisers on fun and engaging AI-infused campaigns.It is hard to describe with words, but you can imagine users generating viral photos with a brand advertiser’s branding that fits the brand image. That gives the uploader a lot of likes and gives the advertiser a lot of exposure.So we continue to experiment with those things. But in any case, we are not relying on advertising for business growth.Grace Shao:On partnerships, I had this idea and I do not know if you are doing anything like this. Would you partner with some consumer-facing chatbots in China to help them with video and visual capabilities?For example, could someone go into a consumer chatbot and call up Meitu’s capabilities? There may also be competition there. How do you view your relationship with these players?Gary Ngan:We are open. In fact, we are already an official partner with WeChat, not on the Xiaochang side, but in another area. I do not remember the exact English name, but basically when you are using the chatbot, you can call up Meitu.Right now, it is still a lighter relationship, almost like traffic redirection. But our goal is to democratize design. Being able to work with more people and enable more people to access that power to express themselves is something we are open to.Grace Shao:That makes sense. It feels like they may not want to put as many resources into this specific use case, and you have the know-how in doing the best video and image editing.Gary Ngan:I would not say they do not have the edge. I think they may just not want to focus on that.Creating these applications requires a lot of focus. It requires the right organizational structure and a laser-sharp focus on trial and error, and on creating the best aesthetic output for users.These may not be the things that larger companies want to invest in. It is important relative to our size, but to them it may be something they do not want to focus on. If they wanted to do it, I think they could.Grace Shao:Let me challenge you a little bit on big tech. In China, ByteDance and Kuaishou clearly have a lot of edge and moat from massive pools of image, visual, and video data. In the West, we have Canva, Adobe, and other global applications. Even Shopify and Alibaba are creating e-commerce staging and design tools.In this big world of competition, or peers if we put it more nicely, how do you see Meitu’s strength? Who are the most relevant competitors that are similar to what you do? And who may look similar on the surface but are not actually doing the same thing?Gary Ngan:We have to separate it into two categories.On the leisure editing side, with the exception of one business unit within ByteDance, there are not many large companies doing that globally. I do not think there is any real large company doing that in the U.S.There are smaller companies, but they are much smaller compared to us.On that side, our edge is really continuing to follow and set the trend for the latest aesthetic standards and what helps users stand out on social media. These are the things we have been doing for more than a decade, and we will continue to excel in them.On the productivity side, there are many companies doing similar things, but taking a much more general approach.For example, Canva and Adobe use one product to satisfy different verticals. Adobe is organized around media: photos, vector diagrams, video, effects, and so on. Canva is one editor trying to fit many situations. Figma is also one app serving many applications.They are design-driven. We, on the other hand, are much more vertical-driven.We are not restricting ourselves to a specific media type. We are saying, within e-commerce, what do you need?You need product photos. You need very short product videos. You need the ability to generate batches and batches of photos. You need to know the latest trend on the e-commerce platform you are selling on. For that particular product, you need to know the selling points. You also need to know the rules of Amazon or Temu and what you need to abide by when selling those products with pictures.All these things are baked into DesignKit.If you are using Canva, I highly doubt it will have a red flag saying, “You should not be using minors in this product photo.” Canva may not even know that you are creating a product photo in the first place.So these are the different focuses we have.In terms of competitors, it is hard to say who is a direct competitor, because at the end of the day, you can use Photoshop, Canva, or our products to create an e-commerce photo. They are all peers, but we take different approaches.If we take a step back, generative AI is still very early. It is 2026 now, but generative AI really only started in earnest late last year for visual use cases. Before then, a lot of generative AI photos still looked AI-generated.Grace Shao:They were quite bad. There might be six toes, or the face was disproportionate.Gary Ngan:Even if there was nothing obviously wrong, you would look at the photo and know it was AI-generated. It did not feel real.Now we are just starting to see things that are harder to distinguish between human-made and AI-made. This is how we can empower the industry and increase efficiency.We are still very early in this market. That is why we are very optimistic and see a lot of opportunities.Grace Shao:A little side note: in 2019, when I was still with CNBC, I covered deepfakes. At the time, there were a lot of deepfake videos of Obama or Zuckerberg. A startup even made a deepfake of me. It was literally just plugging someone else’s face onto my head and body, and nothing really worked.But now, fake images and videos are getting very hard to distinguish with human eyes. How do you view the ethical side? How do you stop misuse of the technology?Gary Ngan:First, we have put in safeguards.For example, on Picchi, if you generate a model of yourself using your own photos, that model cannot be applied to anything other than your face. If we detect that it is not your face, we will not allow you to apply that model to another face.In some of our generation applications, we have also put in safeguards around certain words, such as violence or pornographic images. You cannot generate those using our applications.So there are safeguards that we put in place. Obviously, we can only do so much.One thing that makes it slightly easier for us is that we organize our applications into different verticals. Users come into our applications with a very strong intent. They know they are creating e-commerce photos, for example.Instead of giving them a general chatbot where any random person can come up with a random idea like putting their face onto the President of the United States, it is harder to imagine someone using DesignKit to run a prompt like that.Organizing into different verticals also helps us mitigate the risk a little bit.Grace Shao:The last area I want to talk about is globalization and your global strategy. Meitu is globally available. It is interesting because, as you said, you focus on each vertical, and you have not done a big splashy general marketing push. It also feels like that is true geographically. You are in Southeast Asia, Japan, Korea, Europe, the U.S., and so on.Help us understand global scaling. What have been the challenges? How have you done it successfully? And how do international users from different regions behave differently from users in China?Gary Ngan:I will answer the second part first. Users in different regions all behave very differently.Grace Shao:Give me all the stereotypes.Gary Ngan:Not stereotypes, but I will give you one example.We were doing a user focus group in the UK and spoke to a male influencer. He said, “Your app can edit my jawline? That is incredible. I would totally pay for it. But I do not think it is a good idea to smooth out my skin.”Grace Shao:That is interesting. So it is not okay to pretend you have better skin, but it is totally okay to have a chiseled jawline?Gary Ngan:He did not mention whether it was ethical or not. That was just his feedback, word for word.The challenge, or the interesting thing, is that we have to really listen to what users want in those markets. Different geographic locations need the right mix of features and marketing campaigns.I will give you another simple example. A few years ago, we were looking at Lunar New Year. Koreans also celebrate Lunar New Year, and in China Lunar New Year is a festival where we get a lot of usage.We had launched features in China that were very popular that year, but in Korea there was no uptick. Later, when we had local Korean colleagues helping us run marketing campaigns there, they told us that Koreans generally celebrate Lunar New Year with white clothing and a white theme, while Chinese people celebrate with red.Our Chinese marketing team was surprised, because in China, white is usually associated with funerals. It did not register.That example tells us there are many things we need to immerse ourselves in culturally to understand how people behave, what they care about, and what the standards are.We cannot stereotype anything. Every place and every person behaves very differently.That is the biggest challenge, but also the biggest opportunity.Now we are setting up offices in different parts of the world. We are sending product managers overseas regularly to do more focus groups and, more importantly, to experience the lives of the users they are trying to serve.In the past, we relied too much on consultants, reports, or reading online. That is not enough anymore. We are fixing that, and I think we are making progress in some countries.Fingers crossed, we will continue to grow bigger in Western markets.One other tailwind that has helped us is TikTok and K-pop. Back in the day, editing a photo seemed socially unacceptable to a certain extent. But with TikTok, people are more relaxed about filters being applied and playing around with your face. It is no longer as taboo in many Western countries.The rise of K-pop is also influencing cosmetic styles, and that becomes a segue for us to try different things in Western markets.There are very interesting things happening. But the most important thing is for us to really understand, live, and breathe those cultures so we can create things users want.Grace Shao:That is meaningful, and it feels important for a new generation of Chinese companies going global. Localization cannot just be reading headlines or high-level reports. You have to understand the culture, because culture influences the business.To wrap up, I really appreciate your time. My last two questions: first, what is the biggest misconception investors currently have about Meitu’s business?Gary Ngan:One of the biggest misconceptions is that general models are going to destroy everything, and that there is no place for AI applications.We think that is quite unlikely on the visual side. I am not sure about the language side, but on the visual side, aesthetic standards are very subjective and personalized, and a lot of controllability is needed.Different verticals have different interpretations of the same words. So the way models are trained and organized, even with agents, makes it unlikely that a one-size-fits-all general model can satisfy all verticals.Every vertical has its own workflow and standards. AI application companies are very important in making those adjustments and optimizing workflows for users.That is the biggest misconception.Grace Shao:I agree with that. We are seeing more of that realization in the market now. You have strong vertical use-case AI-native companies coming through, like Harvey. I have also met companies where former investors are building equity analyst research tools.You can say generic GPT can be used for research very easily. But to your point, these teams know the niche use case. They know the process, the standard, and the workflow better than anyone else. Even if the TAM is small, it can be big enough for their business.The last question I ask everyone on the podcast is: what is one differentiated view you hold? Something that is a bit against consensus.Gary Ngan:Is it related to the company or the industry?Grace Shao:It could be anything. Usually people answer about their topic, but it can be anything.Gary Ngan:I think life expectancy will be a lot longer than we think today for our generation.Grace Shao:So we are going to live to 150, thanks to Bryan Johnson’s experiments?Gary Ngan:Possibly. Then there will be more time.There is a lot of advancement in AI. It speeds up many pharmaceutical processes. You can run different trials much faster and understand the underlying issues more efficiently than before.And with more time, there is more time for us to create more art.Grace Shao:And live a healthier life. Although right now, anyone working in AI knows AI never sleeps, and I think we are all working more than ever.But thank you so much. That is definitely a differentiated view. I really appreciate your insights and your sharing today. Thanks again, Gary.Gary Ngan:Thank you so much, Grace, for this opportunity. Really nice talking to you.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Paul Triolo on Chinese labs makings chips, SMIC, Huawei and the importance of AI governance 13.07.2026 1ώ 25λ

  Hi all, we’re back with the podcast. This is the perfect time for Paul Triolo to join us as he gives us a preview of the upcoming WAIC and walks us through some of the top-of-mind questions we have about China’s AI space right now. I do want to apologize for a bit of the echoing in the background; lesson learned to always use headphones going forward.In this episode, I speak with Paul Triolo, partner at DGA-Albright Stonebridge Group, about how to think clearly about China’s semiconductor ecosystem, Huawei’s role in the domestic AI stack, and whether U.S. export controls are actually working as intended.We start with the latest debate around restricting foreign access to advanced Chinese AI models, before moving into the semiconductor stack itself: SMIC’s role, capacity bottlenecks, domestic GPU startups, hyperscaler chip efforts, Huawei’s vertical integration, and why software ecosystems like CUDA, CANN, and MindSpore matter just as much as hardware.Paul argues that the usual framing- whether China can “catch up” to Nvidia or TSMC is simply too narrow. The more important story is that export controls have pushed China toward a broader systems-engineering response across chips, tools, packaging, memory, software, and cloud deployment. We also discuss HBM, rare earths, remote-access loopholes, the logic behind Huawei’s roadmap, and why the collateral effects of controls may be larger than policymakers expected.We close on the bigger strategic question: whether the U.S. and China are drifting into an AI race dynamic that raises risks for everyone, and why more direct dialogue — not just more restrictions — may matter most from here. [Paul co-authored a piece here discussing how to navigate the complexities of the U.S.-China AI safety dialog] This is an extremely insight-dense episode, and I hope you enjoy it as much as I did. Thanks again Paul Triolo. Btw, coming up next are a few episodes featuring founders and execs from hot-listed AI, autonomous driving, and spatial intelligence companies.To find the previous episodes of Differentiated Understanding, see here.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.Season two will host a series of guests from analysts, VC investors, builders, researchers, founders, and product managers. For more information on the podcast series, see here.Chapters 00:00 Beijing, Chinese AI models, and why regulators are paying attention04:43 Why AI labs are moving into chip design08:12 SMIC’s role and the fight for domestic chip capacity17:56 Huawei’s capabilities and why it became the center of the conversation26:53 CUDA, CANN, and whether export controls really worked38:24 Can Huawei’s software stack win developer mindshare?51:10 Why China can still progress despite compute constraints57:07 The AI race narrative and why Paul is skeptical of it1:07:46 Why governments still lack the technical capacity to respond1:08:42 Will frontier AI labs eventually be nationalized?1:13:59 The risks of zero-sum U.S.-China AI policy1:20:46 Why more direct U.S.-China dialogue mattersTranscript (AI-generated for reference only)Grace Shao (00:00)Hi, Paul. Thank you so much for joining. Really excited to have you on. And I feel like you’re the perfect person for a few of the questions I have prepared in the beginning of the podcast before we get into the actual topic. There are so many things happening, and I can’t keep up with like Twitter these days. So no, so supposedly Reuters reporting saying Beijing is restricting foreign users in accessing Chinese models. Like, what is that all about? What’s your view on that?Paul Triolo (00:15)It’s hard to keep up. Yeah, I think in the wake of the fable mythos fiasco, if you will, in the US and the capabilities of these advanced models getting really good and getting into areas like cybersecurity and biosecurity, I think it’s not surprising that the Chinese government is at least considering what to do about open-weight more properly models that are that are that are reaching sort of frontier level capabilities, particularly DeepSeek and Zhipu and then even more recently, you know, Meituan and others. I think my sense is this is just preliminary discussions with the labs about this issue because, you know, even in the US, there’s lots of confusion about what the government’s role should be here in determining when models are released and under what circumstances and how do you measure capabilities. even though the US has been thinking about this for a while, and I’m sure that’s to some degree that’s happening in China, there’s no general agreement on how do you do this. And these companies in China, as you know, are all commercial companies that just like in the US are under a lot of pressure to continue to put out models. and so I think I would not read too much into that. I think that’s it’s clear that the Chinese government is Trying to figure out what to do with about this, but I don’t think they’ve reached any conclusion about which models to control and how to control them. they’re they’re learning from the companies. They’re probably going out and saying, you know, how do companies themselves evaluate these models internally, in terms of capabilities that could be of concern? and what should the Chinese government eventually do? I mean, they’ll they’ll they’ll do something eventually, but I think we’re in the early stages still. as a result of theGrace Shao (01:57)Yeah, for sure. I think you know, Zhipu, Minimax, these companies are public listed, like they actually face, you know, just shareholder pressure as well. So it but the one thing, the nuance is Chinese companies usually are a bit more prepared or aware of potential regulatory, I guess, involvement. so yeah, let’s let’s wait and see. Because I read this and I was like, this seems bit counterintuitive, frankly, to the model’s going forward. I think like the deadline was a little out in front here. I think it’sPaul Triolo (02:49)Right. I think that the headline was a little out in front here. I think it’s clear that there’s concern as these models become more about what to do about tiered releasing. but this is a more general discussion I think that’s happening. it doesn’t surpr it to people who’ve been following this sector for a long time, the idea that we would be here at this moment, you know, was not surprising. The problem government gov the ability of governments to keep up with the pace of development of the technology is just clearly here, it’s it’s woefully inadequate to the moment because you know within these large AI labs, and I’m now calling DeepSeek and Zhipu, along with Anthropic and OpenAI, you know, the top four global frontier AI labs. you know, th researchers n understand these issues and they’re they’re really concerned about this because P particularly things like recursive self improvement, which is which means models are basically training the not training themselves, but they’re they’re optimizing some of the orchestration or the harnessing that the sort of platforms that these things operate on. You know, that’s been a that’s a growing concern because they’re you know, the models themselves now are able to improve the overall ecosystem without human intervention. Right. and that people miss that, I think, in the in the US with all the fable mythos kerfluffle. The Anthropic released a blog that talked about this, that recursive self-improvement is now sort of part of the landscape. And so that’s also I think part of the concern including in within the Chinese government about okay, you know, Chinese labs are getting pretty good. what should the government how should the government think aboutGrace Shao (04:43)Yeah, definitely. And I think, you know, in China, usually the regulators and the industry actually work pretty close together. so hopefully, you know, people can kind of regulators can keep up, will hopefully catch up on understanding technology a bit faster and better. Okay, so another quick commentary on what is happening in the news. Supposedly DeepSeek and Kai AI are going out and making their own chips. What is happening? What is your high level view on this?Paul Triolo (05:08)Everybody’s doing it. Now, you know, in it this is not surprising. US, of course, some of the hyperscalers and more of the hyperscalers like Google and AWS have long d determined that it would be useful have specially designed ASICs, application-specific integrated circuits that are optimized for running certain workloads in their cloud. and you know and optimi na and now optimized for models specif specifically for you know large advanced models for which general purpose GPUs, which is what NVIDIA and AMD produce, may be, you know, may be suboptimal. but this is a complicated issue because to do semiconductor design, you know, this is a whole nother thing than building models. And so for both Zhipu and DeepSeek, you know, this requires building a team of design so some semiconductor design engineers, right? Who now they’re not they’re not a lot of these guys laying around that are not gainfully employed, particularly for designing really sophisticated chips here. So I think it’s not surprising that they want to do this, but I would be again sort of a little bit skeptical that they’re gonna do be able to do this in the you within the next year even. You have to build a team. It’s expensive. use you have to get advanced semiconductor design tools, electronic design automation tools. and then you have to begin figuring out where you’re gonna manufacture these, right? And in China, of course, as we know, because of export controls, these companies are likely gonna have to use SMIC, the domestic foundry. So they’ll be competing with all the other players, all the other GP GPU designers, the general purpose GPU designers like Biren and More Threads, a of these companies that as you know have gone public recently, they’re all vying for this limited capacity at SMIC to manufacture these advanced designs at say let’s say seven nanometers, which is a feature size of these chips. so I think it’s really interesting that they’re gonna do this, but it’s gonna be a challenge on two fronts. One is you know, assembling a team Sustaining that team over time, you know, this is expensive. It’s very expensive. Lenovo tried to do this, for example, at one point. They were considering getting into semiconductor design, but they determined it was it was too expensive and it was going to be a long term drain potentially if you know, depending on the success of these teams. and so, you know, I think if anybody can do it, I mean DeepSeek seems to already have a lot of expertise on hardware and understand the hardware really well. But it’s it but semiconductor design is another whole nother discipline, if you will. And then in China it comes with the added constraint that you have to you’re sort of stuck with SMIC and you know, maybe Huahong down the road will have sub some kind of seven nanometer process. but it’s you know it’s it’s a tricky thing. But again, the trend in the industry is to do this. so in the US you have the open AI, I think just recently Anthropic are both also considering you know designing their own semiGrace Shao (08:12)Yeah, they have partnerships as well with other vendors. So help me understand. I think twofolds. One is what is SMIC’s role in China and what is their biggest bottleneck? For them to actually churn out better chips. One thing you said is capacity, other thing you said is access to certain technology, you know, instruments, machinery. The other part of the question, if you can fold into it, is who are the actual players? So, you know, DeepSeek and Zhipu wants to build their own create their own chips, but Baba’s in this, you know, Baidu’s in this. There’s a lot of big tech trying to create their own chips as well. How do we understand all their relationships in the ecosystem?Paul Triolo (09:06)Wow, okay, you got a lot a lot in there to so let’s just let’s just look at the demand side, then we can look at the supply side. So the demand side, as I started to allude to there, is pretty heavy, right? So you have Huawei, right? They’re designing the ascends, and those are all at the at least the most advanced node, which is this horrible technical term for the process that is used to manufacture these at SMIC. For example, here. And so Huawei has a lot of demand. And past, Huawei was given most of the capacity at SMIC, just because a year ago or two years ago there weren’t as many other players. Now there are. Now it’s more complicated. But certainly the Chinese government probably also heavily weighed in to have SMIC prioritize Huawei production, both for their smartphone, their Kirin smartphone, and for the Ascend processors for AI. But now in the last year, we have all we have a we have two s two additional sets of companies vying for this capacity at SMIC. One is the G the sort of GPU makers in China, the startups. And this is Biren, Moore Threads, SoftGo, Inflame, Iluvatar. You know, there’s at least there’s a couple more too, but those five are sort of the top companies. And some of those companies were started by engineers from Nvidia and AMD. And so they their designs are really advanced and they are they’re more compatible with the NVIDIA ecosystem, et cetera, et cetera. So they’re all now producing GPUs at SMIC, and there’s allocation issues. I when I was in China recently, I heard that one of those companies had been promised certain number of wafers at SMIC, but then one of the big hyperscalers would come in and offered more money. And so SMIC had said, Okay, well, we’re gonna reduce your allocation, right? So there’s a lot of fighting for that. Th and then the other group that you’ve just mentioned here is are the companies like Alibaba, the hyperscalers, and then now the model developers like DeepSeek and Zhipu that are also doing their own designs. And I think again, as you noted, Baidu and Tencent and Alibaba are more are farther along on this. They have semiconductor design teams which are already producing, in the case of Alibaba, and in the case of Tencent, they have dedicated ASICs. They’ve been doing this for a while, sort of under the radar. Tencent’s a very capable company and has been they really good at this, right? It turns out, but they don’t they don’t they’re very low key on this. and then Baidu of course with Kunlunxin, which is also gonna go public on the Hong Kong market. So the other part of this also is that all these companies of course need capital To function and to do all these designs and to hire these engineers and to and to do all this stuff. So all the in the last year we’ve seen these companies tap into capital markets, particularly, you know, the GPU makers, the new GPU startups have gone public in Hong Kong and in Shanghai. And then you know, we’ve seen DeepSeek, of course, raise seven billion in funding from a variety of sources, and part of that will go to presumably the building up a design team and doing Designing their own chips. and then of course Zhipu has gone went public and its stock is crazy high on the if you look at the valuation on Hong Kong. So it’s a new a new game where Chinese companies are playing this game of d you know, designing their own chips. and then of course they’re all competing for this capacity at SMIC. So now we can turn SMIC, right? So SMIC is this crazy company, right, that for a long time was you know was sort of under radar. but they are under heavy US export controls. So that started in around twenty eighteen, twenty nineteen, when they were they had actually ordered a very advanced lithography machine from ASML, which produces all of the advanced lithography machines, extreme ultraviolet lithography, UV lithography. So they were denied that. They actually ordered it. And then they and then the Dutch government, under pressure from the US government and Wassenaar, which is this interagency or intercountry group, this multilateral group, pulled that license from them. So for the last whatever, six years, SMIC has been using its deep ultraviolet lithography, which is the second best, but pretty good. They’ve been pushing their suite of DUV machines to the utmost limits to try to get to these lower and more advanced nodes. And seven nanometers is sort of the limit, seven and maybe five It’s complicated. Some layers of these semiconductors can be but they don’t all have to be at the most advanced levels. but they’ve been doing something that nobody else in the world has been has done. Now in Taiwan, they did use some for example, that SMIC is using, but they that when they had a access to the more advanced lithography, they went to that because it’s it’s it the throughput is faster. And the yields are better. So SMIC is trying to do something that you know that nobody, no but no other company would have to do because of US export controls. And that means pushing these lithography machines to their limits. they’re having obviously there’s a they’re being successful, but there’s a limit to sort of the yield too that they can do. So for example, the AI semiconductors are much more complicated than for a smartphone handset. So for the smartphone handset, they yield. you know, say ninety over ninety percent. But for the GPUs or the NPUs as actually as Huawei is using, the yields are much lower because these are very complicated and dense chips, right? And so using some of these advanced techniques just it’s just hard to do, right? And it’s and you end up with not as many use useful chips at the end because the yields you know, you’re you can’t do it because you’re you’re you’re sort of pushing the machines to their to their the limit of capabilities. So but this neck but over the next six months it looks like they’re gonna bring online SMIC is gonna bring online more capacity at these advanced nodes, maybe double capacity, because obviously they there’s so much demand for this for this these chips in China that SMIC is responding to this demand and is putting in place more production lines in Shanghai, by the way, at SMIC South to produce to meet the demand for all of these. AI primarily not just AI but mostly AI optimized hardware because of these because of those three batches of companies. You know Huawei is sort of a batch in its in its own right. But then they the GPU makers, the startups and then the ASIC makers. So the challenge then is who decides who gets the capacity, right? and as I said, there I’ve heard a lot of anecdotal you know, chatter in China about that. It’s very complicated. The government obviously weighs in to favor certain companies, but then you know, there’s a tremendous amount of competition. And then finally, the other piece of this not just the logic, if you will, the sort process or die. It’s also memory, right? So for more for the for the AI hardware, high bandwidth memory is a really important part of that because These advanced GPUs are packaged with lots of memory on the on the actual package, in within the actual package, co-packaged, if you will, with the logic. and there, of course, US export controls again have affected CXMT in particular, this case ChangXin Memory, is in Hafei and other places in Beijing. I just saw a big fab in Beijing when was there. and so there but again, it turns out is it to of the export control restrictions on memory is not quite as hard as it is for logic, although it’s not easy. and so CXMT is also ramping up its production of high bandwidth memory, which would then be packaged with those with those dyes. Huawei in particular stockpiled a lot of HBM in twenty four before the US controls were put in place from what Samsung and SK Heino, some of the South Korean producers which are leading In the production of high-bandwidth memory. So anyway so it’s a combination of several bottlenecks that the Chinese domestic semiconductor industry is trying to overcome to meet this demand for these for these AI this AI optimized hardware, whether it’s the Ascend series from Huawei or these other GPUs or these other ASICs that are that are now being designed by the hypo hyperscalers and the model developers like Zhipu andGrace Shao (17:56)I appreciate that context. I think for me, like I’ve been reading about SMIC for a long time, but that just really clarifies exactly their role in the ecosystem. And it’s interesting to kind of see how they’re prioritizing certain companies over others. but anyway, I want to double click on Huawei. I think it gets the most heat, you know? It obviously is not And it’s a very interesting company because it’s not just such like it’s not only just like you know pushing out their own models, but they’re a foundry, they’re they’re a chipmaker, they’re a designer, but they’re a bit of everything, they their own hardware. And then they also have obviously then CANN and then MindSpore that trying to compete on the ecosystem side with NVIDIA. So walk us through just Huawei’s capabilities, Huawei’s competitive edge. And then why Huawei was put on the entity list and got so much heat over the last couple of years.Paul Triolo (18:44)My God, great question. Great question. You’re really asking good questions here. But you know, there’s some, and I’ve written so much on this. So look, Huawei, if you remember, was primarily a telecommunications equipment company. the in the 2000, 20, that, you know, up until say 2018, 2019. And then Huawei got into the handset business too, right? They built they started building smartphones, they were ramping up. you know, to they were competing and out competing in some sense Samsung and Apple. And then the US put Huawei on the entity list in twenty nineteen, May of twenty nineteen. I still remember where I was when they were when I heard they were put on the entity list. It’s like when I when I remember where I was when shot. and then they were and then they were the even more critical in twenty, they the US added this so called foreign direct product rule, which meant that not manufacture its designs at TSMC basically. And I remember being in at Huawei in 2019 when I toured the their headquarters in Shenzhen and they showed very proudly all of the semiconductor designs they were doing at the most very advanced notes for all of their product lines. But at the time those were you know there was more on telecommunications equipment and also on cloud and on you know servers for clouds, the Kunpeng series of chips, for example. So anyway, so Huawei and the secret there that still important is that Huawei had spent a lot of effort to develop their that design team for those semiconductors, right? So I mentioned earlier how hard it is to do that. Huawei had over the about a ten year period had developed this arm of the company which was designing by the by the twenty time frame, was designing cutting edge chips, you know, on par with like Qualcomm and Nvidia even and some of the other areas. And it was getting better and better before the US basically you know w by putting Huawei on the entity list, HiSilicon was also there. So then High Silicon could not use TSMC. But in the process, Hi Silicon of doing some generations of chips at TSMC, Hi Silicon gained a lot of knowledge about you know both semiconductor design and how to use those complicated tools, those EDA tools, and how to do manufacturing. Because when you work with when you’re doing design and you’re working with a fab like TSMC, you’ll learn a lot about how that happens. So high silicon is sort of the secret weapon and if you will of Huawei. And so after the US controls, the Huawei kept high silicon designing. The designers kept designing Even though they didn’t have a place to manufacture yet, right? And eventually SMIC, working with Huawei figured out how to do all these optimizations of their existing equipment, this DUV equipment, to allow Huawei to manufacture SMIC, even though they weren’t the most advanced process, they were still pretty good, right? And Huawei made all these innovations in terms of overcoming some of the that not being not having access to the latest and greatest tools meant, which was, you know, for things like power consumption and other things. They did they designed around this, right? So that’s the other sort of this is a theme that you’ll that I’m sure you’re familiar with is, you know, the force Chinese companies to do different things and optimize things. This is what happened with DeepSeek, right? And other Chinese companies that don’t have access to all the latest and greatest Nvidia chips. So same thing with Huawei, they figured out how to design around this. Now the other thing they did, of course, which I don’t think mentioned was software, right? Because the US controls, for example, restricted access to Google Mobile Services for their handsets, Huawei invent had to invent Harmony, the HarmonyOS, right? So Huawei had to become a software company, right? Which they didn’t really want to do. Arguably, I talked to the senior Huawei officials and they were like, wow, you know, if we could use Android, why would we go to all the trouble inventing having to invest in and build A whole new operating system for which it doesn’t generate any revenue, right? It’s it’s just it’s sort of a cost center for Huawei. But they had to do it because they were under pressure. And so as part of that process, it’s important to understand that because now we get to the AI stack that you mentioned, the AI era. So Huawei has as a result of US export controls and having to develop Harmony, they have now known sort of the process of how to develop a software ecosystem. and how to get developers to use it, right? Although again with AI it’s it’s harder, right? So you mentioned CANN, the Compute Architecture for Neural Networks, which is Huawei’s equivalent of CUDA, which is the sort of developer and software developer environment that’s critical for NVIDIA. So yes, Huawei is trying to now develop as it did with HarmonyOS. And that was a long process by the way, a long and hard process. It wasn’t easy to do. They had to they and still, right, still China s smartphone companies still use Android, right? They don’t all use Harmony. But Harmony is sort of a cross device thing that you know goes for automobiles. You can use it on your car for the Huawei the Huawei invested EVs. So it’s so it’s a it’s a pretty good system. I’ve seen it when you go with your phone to your car, it’s all seamless, et cetera, et cetera. So anyway, so Huawei knows how to do software development in a complex hardware environment now. And so for AI, which is harder, they’re doing yes, they’re doing can is something that they’re working with. And then also MindSpore is sort of the equivalent of like PyTorch and it’s an it’s a development environment that AI developers use. And so they’ve gone a come a long way on that. In 2020, 2022, 2023, you know, because Chinese AI companies could still use Nvidia, they you know nobody was using Can and Huawei. But now DeepSeek and Zhipu and even Meituan, which looks like they have their million perimeter trillion parameter model on 50,000 ascends. This is something that Meituan has complained. And then now we have Minimax saying they’re going to do a 2.7 trillion parameter model. They’re all working with Huawei to optimize some of the aspects of that software development environment. Now it’s complicated because there’s training and there’s inference. And so there’s different needs for each of those in terms of development. But the again, the US export controls have forced the Chinese model developers to work very closely with Huawei because Huawei is the main alternative, right? And the Chinese government, of course, has been encouraging this. And so, you know, now we’re in the situation where the development environment around the Huawei and the SENS and CAN and MindSpore is better, arguably, than it was even a year ago. When I was at the World AI conference. last year in Shanghai, I talked to a lot of hosting, you know, various capabilities using Huawei Sense and they said that the Huawei system was hard to work with. They would tell me, they wouldn’t tell me this, you know, they I didn’t want to be quoted on this, but they said, you know, they were being told to use Huawei hardware, but it was hard for them to offer the kinds of services they were offering with Huawei hardware the same the same caliber of as with NVIDIA. But now I think that gap is closing. It’s not like Everybody in China’s all the AI developers are rushing to Huawei. but there’s still a complicated mix of both NVIDIA hardware and as we’ll see now, they’ve they’re the Chinese government is allowing I think 10 companies to buy these H200 GPUs, which we should talk about. anyway, so it’s a very complicated and heterogeneous compute environment in China. But one thing we can say with some certainty is the Huawei because of the export controls and because they’re closely with. DeepSeek which is very good at programming the hardware, for example, the that environment is at a stage where probably it wouldn’t have been without the export controls. And so we’re in you know, it’s it’s it’s getting better and better. and because companies are gonna have to use it at some point, they’re sort of d deciding, like DeepSeek is deciding, well, we better put a lot of effort into helping optimize that development.Grace Shao (26:53)I think I agree with you and what I’ve been hearing on the ground as well. A lot of the developers saying like if they had a choice, they wouldn’t really leave CUDA just so much better. But if they don’t have a choice, it’s kind of like damn it, I’ll have to just try to learn how to use this. And even if it’s not as good, like we’ll try it’s also like a chicken egg thing, the more developers on it. The better the s the software and the system. But I want to play devil’s advocate here. Like obviously, we all heard the Dario and Jensen like interview. Right. So like we don’t have to get into the details of that. But part of the argument. But part of, you know, what you just talked about was like, you know, high silicon kind of got shafted. They couldn’t get access to certain machinery. You know, obviously, right now we can say that. Objectively, factually, Chinese chips are probably not as good as the leading chips globally. So thus some may argue exp export control worked, right? Like so what’s your view on that?Paul Triolo (28:02)Wow. Okay, that’s a that’s a rather large topic. So it sort of depends on what you mean by work. so look the original goal of the export controls as sort of articulated in the you know federal register notice in October twenty two was originally related to s you know to sort of military other sort of nefarious end uses of right? but the real driving force, if you will, was really the this idea slowing Chinese companies’ ability to develop frontier models down so that the US would get to some advanced level of AI first, right? And so if you just look at that and you look at say Zhipu releasing GLM five point two, that’s not as quite as good as fable or mythos, but it’s pretty good, right? And it’s the gap between le the le the leading models from anthropic and openai and the leading models from Zhipu and DeepSeek and other Chinese companies and Alibaba in particular and now you know Meituan and even Xiaomi and Minimax, you know, that gap is still there, but it’s not really is it months, is it a couple of months? So if you’re going to argue that the export controls worked, then you know, is the does a two or three month gap even matter now, right? so that’s that’s one way to look at it, right? Now, if you look at it in terms it make did it reduce the ability of Chinese companies in the semiconductor industry to manufacture advanced GPUs, for example, on par with NVIDIA, of course it worked, right? Nobody would argue. that it did that didn’t happen because if you’re gonna restrict exports of GPUs and you’re gonna exp r you know restrict exports of critical tools that are used to manufacture those GPUs, of course you’re gonna you’re gonna slow them down. But then they then you have to say well how has Chinese how has Chinese industry responded to that, right? And what are and what are the costs of that for US companies, right, for example. And then you have to look at what is the retaliation from China to those export controls. So you have to at least look at it, look at the picture more broadly than just, you know, did the US slow down Chinese model development? Arguably there, the jury’s still out on that, right? Because I would argue that they haven’t the slowdown hasn’t really been that significant. If a company like Zhipu, like who had heard of Zhipu like even a year ago, right? if they can release a model like GLM 5.2, okay, wow, that’s a that’s a frontier model, right? It’s it’s matching fable in some benchmarks. Okay, so it’s clearly somewhere near the frontier. How close we can argue about and a lot of that is complicated, depends on the benchmarks you’re using. But in the in the semiconductor industry, then the you have to look at that in a little more depth. One thing that I’ve written quite a bit on of course it’s forced Chinese toolmakers to work With the with SMIC and Huahung and some of the other manufacturers, CXMT and YMTC. And so the overall level of capability, for example, of Chinese, the Chinese semiconductor industry to do stuff domestically has gone way up. So those toolmakers, for example, NARA and AMEC and Piotech, they’re now competing outside China with US companies in a way that was inconceivable in 2022. And so what happened, of course, is As a result of the controls, the US companies had to pull all their people out of those fabs in China. guess what? US competitors, US company competitors from Japan in particular, and also Chinese domestic companies had got access to that equipment. And they learned things from that equipment that they wouldn’t have learned if the US companies had been in control of that equipment. And so that’s one just one of many examples of sort of the way you have to look at this if you’re gonna say, did they work? Because as a result of the controls, the US now US companies now have competitors globally for the in the tool making sector. So for example, NARA and other companies in China have been qualified for TSMC to provide tools to TSMC and to Intel and Micron, right? and so now US companies face bigger competition. And the ability of China’s semiconductor industry to pr to eventually produce more advanced chips has gone way up, right? So because the that semiconductor part is complicated. You know, the idea that the US, example, could use controls to forever keep Chinese companies from developing n capabilities sort of, you know, it’s unrealistic, right? Because th this is a this is an applied science. And so the ar the US argument was this is a choke point that we can stop China from doing, but no, China’s designing around that because there’s many ways to do things, right? There’s there’s more than one way to do to develop a tool, for example. And The industry has pursued many different paths and over the years, some have more commercially viable. Those have been the ones that dominated, but now China is pursuing other ways to do things. And so you know, that’s that so like hu like Huawei, just a quick example. So Huawei just you probably saw a couple those last month, I think. They came out with this Tao scaling idea. And so is the re reduction of feature size is to increase the speed and the and reduce the power consumption of these chips. And so that’s Moore’s law has held for a long time. But now we’re run, you know, the industry is running up against just the limits of physics in that in that regard. We’re down to you know one nanometer, you know, really small feature sizes. And so Huawei is saying, okay, well there’s maybe another way to do that. We can we can use a sort of three dimensional structure here to also to move the components closer together and to reduce the time, the latency between signals going to those components. And so that’s not new in industry. This approach has been used before or you know, people have been looking at this. But Huawei is now putting a lot of effort into the actual tools and the technologies to actually do that at some kind of scale. Now the jury’s still out on when that will happen. They’re saying by 2030, for example, they’ll have a s a feature size that will be a system that will be of like a 1.5 nanometer system. And again, the other thing to remember here is that it’s not now just about feature sizes, it’s about sort of the entire package of the system, right? It’s about the memory, it’s about the interconnections, the optical interconnections between the GPUs, where Huawei, for example, has a lot of knowledge of optical interconnectivity. And so you can’t just you can no longer just look at the individual sort of feature size die to die and then and then determine that you know China is ahead of the US or US is ahead of China. So it’s a more much more complicated calculus. And I’ve written about this quite a bit. But you know, I wrote the I think two years ago I noted that you know that now we were in a different ballgame. It was really systems engineering at a at a higher level that’s going to determine you know the capabilities. And here again, Huawei has some significant advantages. so anyway, so the long worded answer to whether the export controls worked is well, yes, of course they worked at some degree to stop and slow down Chinese industry. But at the same time, they’ve accelerated key parts of that industry. And then finally, I would argue the rare earth issue, which by the way I live every day because we’re trying to help companies overcome some of the issues around many licensing and other things, you know, that has been a huge thing because that was directly responsible in response Gallium, graphite. And then of course in April last year, the controls on heavy rare earths and magnets, right? And so those are still, you know, with us. Just today, you may have seen and yesterday, and Nikkei had a story about Japan, Japanese companies who which have been cut off from rare earths in Jan starting in January. They’re filing with the Tokyo stock exchanges are indicating they’re because they’re running out of these critical materials. And all of that result of the of the US export control regime and China’s response, which is to put in place this very strict licensing regime around rare earths, the way, are key inputs for the semiconductor industry too, right? So yttrium, for example, is used to line etching chambers. and most of all those machines I mentioned, DUV, UV, they all use lots of rare magnets for various purposes. every ch semiconductor produced in the world, virtually every one, is touched by a plasma, which is this gaseous, you know, material that’s controlled by Chinese rare earth magnets, and the chambers where that plasma is contained are lined with Chinese rare earths materials. So in other words, the export the US ha put in place have resulted in this very serious response China. That we still are in the middle of. We don’t know how it’s going to come out, but it’s already had a huge impact on the entire supply chain for the semiconductor industry. And not just semiconductors, but of course and power tools and any industry that uses these materials. So anyway, so the disruptions that caused by that are huge. And so when you’re looking at the cost, so we’re looking at the costs and benefits. Did the US slow China’s AI development? Yes, degree, but Jury’s still out on how much. And then if you look at all the collateral damage that those controls cost, you know, those are stacking up and there’s no end in sight right now. So that’s but my view is always, you know, you can you can I agree that the controls work to some degree, but then the question is, you know, what was China’s response both from an from an industrial point of view in terms of working around the controls, and then what was the collateral damage created by the controls and that is that is considerable, I can tell you.Grace Shao (38:24)Paul, I love interviewing guests like you because I was gonna follow up with like HBM in the whole picture, how that affects it. You already answered. I was gonna ask you about inferencing versus training on Huawei chips. You answered it. I love you give the full picture. but I wanna ask, what is it like you talked about collateral damage and how these like industries kind of came out because of export controls? Now, how do we understand actually potentially CUDA? I sorry, not CUDA can. Taking some market share, I wouldn’t say lead at all, but some market share away from CUDA and potentially courting more developers globally, maybe beyond just China. How does that new ecosystem and operating system meet work? Because I would challenge and say harmony at this point is still nowhere close to being a dominant operating system, right? So despite you making the point that they obviously had to go around it and create harmony and it does exist and suffice for their own hardware ecosystem. It’s not a leader. How do I understand that?Paul Triolo (39:23)Yeah, yeah, that’s a great question. That’s a great, great question. So, you know, this is a this is an older question remember, you know, the China didn’t the Chin there was no Chinese operating system, you know, for j like for PCs back in the day, remember? so we had things like Red Hat Linux, you know, or Red Flag Linux, right? Which was the which was a sort of source Chinese version of Linux that was touted as gonna you know, that was gonna be sort of the Chinese version of Windows because of course China has been dependent on Windows for a long time and still is to some degree, right? And so this big the big this issue of sort does how does how does China how does China develop alternatives to existing dominant software ecosystems like Windows or like Android CUDA. You know, is a is a is a really good question. And it’s and it’s sort of it’s a complicated issue because each of those has a different a different dynamic there. And it’s and it turns out to be really hard, right? Because developers and I know this from installing CUDA software environment on my home computer where I have a I run an RTX 4090 NVIDIA GPU, which is export controlled to China, but I wanted Seek. a deep seek model on my home my home system and I had to install all of this development environment which is very complicated which included you know PyTorch and CUDA and all these things, right? And so when you’re a developer and you’ve been working with all of these things for many years, the idea and somebody tells you, you’re gonna have to now switch over to this other system, which you don’t know, and you don’t know the limitations and the and the strengths and the weaknesses of that be like, Like really? Do I have to do that? You’re not gonna want to do that. You’re gonna resist, right? and so this and same with Nope when Chinese companies were using Windows and somebody said, Hey, here’s red flag Linux, which of course wasn’t very good. and you by the way, you can’t run all your Windows applications under Red Flag Linux. so you’re gonna have to run, you know, weird open source versions of all of all your favorite programs. Again, you know, I did that for a while. I actually Linux exclusively for a while, but then I ended up coming back to because you know, there was certain things I couldn’t do. so same thing here and same thing with Android and Harmony. So it’s a it’s a but as I said, when like when Huawei first started on Harmony, you know, they had a hard time convincing developers in China to use Harmony. But now you know I think that process is pretty far along and other I re just recently you know other companies are starting to use Harmony and so Event it depend and again it depends on their business model. If you’re s a Xiaomi and you want to sell handsets outside of China, you’re probably gonna go with Android because you can still use Google Mobile Services, right? I mean it was really a d a devilishly clever thing for the for the administr for the for the Trump administration to control access to Google Mobile Services because that really killed Huawei’s business China. And that was a key source of revenue, by the way. So that was not an accident, right? Like wh at one level it was like why should they do that, right? It’s not military technology. It’s it’s it’s you know, YouTube and Gmail, right? But the reason was they really wanted to kill Huawei’s handset business. And so that’s why they targeted that. But other Chinese companies can still use Google Mobile Services. So Huawei in that case is operating in a in a in an environment where Android is still out there, they haven’t Android is still available in China. So they have a they’re competing against they’re still competing against Android. Now CAN, it’s tricky here because in addition to Can, as I mentioned, those other GPU companies like Biren and others, their develop they have their own development environments. And those development environments are more compatible with And then in addition to that, Huawei is trying to make CAN and the whole environment more compatible. with CUDA. So the idea is that you know the difference between the two. If it was here three years ago, now the difference you know, is less. And so it that willingness of the developers to move to environment easier as you sort of reduce the differences. And so and it’s hard to gauge exactly where that is, right? Because each and each company is different. So DeepSeek, example, I is different in the sense that those guys were programming the hardware directly, right? So if you don’t if you if you are really good and not that many have engineers that can do this, you don’t need CUDA. You can program the hardware directly, right? You CUDA is sort of this intermediate layer that makes it easier. It’s a bunch of libraries and it makes it easier developers to train you know use the training environment. But if you know how to program the hardware directly you don’t need CUDA. So anyway, DeepSeek is sort of unique in that they were really good at the hardware. and so that’s why it’s important that they’re working with Huawei because they understand you know the sort low-level way that these systems all work together so they can help Huawei to improve the ability of the capability of Huawei’s hardware development environment to more to be with CUDA. And so I think we’re in the process of having that happen What’s probably gonna happen in China is gonna there’s gonna be a sort of s system where Huawei will be and CANN will be used more for inference on the inference side to inference and some and CUDA and NVIDIA will still be used to some degree on the training side. Because remember, it’s complicated. Right now, Chinese companies can still, for example, use remote access to services like in Japan and Southeast other places. to train their models. And so they can continue to use the NVIDIA development environment for that, right? And then and then when you get to inference, they can then use Ascend and they can run that on there and they can they can optimize using CAN to run on the on those on the on the for on the inference side. So we’re in this sort of a weird world where you know there’s the developers haven’t all switched over to Huawei and they still don’t really want to. But more and are there’s more effort And ease that transition CANN with CUDA. And so where we exactly we are in that is hard is hard on any given day is hard to say. But clearly, as you noted earlier and I and I tried to stress, the problem is that for the long term, the Chinese government and these companies don’t know what the US policy is here. So we just saw that hundreds, you know, maybe two hundred thousand H200s will probably be approved by government. For companies like ByteDance and Alibaba and Tencent and others to buy, right? Okay, so they buy those. They can use those. Those are really good for inference. They can just, know, they can they have a lot of demand for their for their services. They can they can throw those in and they can be used for inference. They can also be used for training if you know what you’re doing, right? You can tie a lot of those together. but what next, right? So what is the US government’s policy? basically, under the influence of Jensen Huang and agreed to stop. Forcing NVIDIA to downgrade their chips for a set sale to China. And so Trump said, okay, you can we’ll allow them to sell, you know, not the cutting edge, but something a couple generations behind the cutting edge. So hence the H200 class GPUs. But what’s next? So if you’re if you’re a Chinese company, you can’t count. Any other in the world doing AI design can say, okay, I’m gonna, I’m gonna, I’m gonna upgrade my cluster from H200s to Blackwell, and then I’m gonna upgrade to Vera Rubin, which is the next one, and then I’m gonna upgrade to Feynman, right? So there’s a roadmap of updating your hardware cluster. China, you know, what’s what comes after the H200s? So therefore the pressure is to and this is why Huawei eventually issued a roadmap, right? Huawei had never done a roadmap for any of this, but now Huawei, because dynamic, had to come up with a roadmap. And so that’s why they have the Ascend 950 you know, the nine the nine twenty and nine fifty. So now they have a roadmap out to twenty thirty or tw twenty thirty one that’s their roadmap for upgrading their the domestic processors. So if you’re a Chinese company, like ByteDance or like you know Alibaba or Tencent, all the leading players, you have to figure out a very complicated equation which is how do I keep my core developers who are using CUDA happy, using some hardware, either in China and then how do I gradually transition to using domestic hardware for some workloads, right? And again, these companies can they can run different workloads on different systems depending on what the need is. and so they’re and then at the same time, you know, maybe they can get some GPUs from REN or you know Sofco or some of the other smaller players and run those are primarily inference workloads. And so but they can but those are those are really good, you know, those are very, very capable. GPUs. So they can run some stuff on those and experiment with those. And those are gonna be easier because those are more compatible with the Nvidia ecosystem. So anyway, so have a very complicated hardware environment to navigate compared to Western companies. You know, like OpenAI can just keep its clusters depending on you know how many GPUs it can Nvidia and AMD. so it’s it’s it’s a very interesting and heterogeneous situation here. Where it’s different than Harmony because Harmony is, you know, it’s still developing the developers develop apps to run on Harmony, right? And so you have you have that piece. it’s like a it’s there’s inference and there’s training and there’s a lot of different things going on here. there’s runtime stuff that you’re doing, there’s harnesses, which are the ecosystem around these models that make them capable. so AI development environment is much more complicated than har than for a mo just a mobile operating system. Will. But again, the export controls and the uncertainty of policy really they’re there, right? And you know nobody has said that eventually the US will allow black wells to be exported to China, for But we’re still in this weird Chinese companies can and access those restricted semiconductors outside of China. They can they can run training workloads in Japan, right? And so that loophole may be may or may not be closed over the next year or so. but in the meantime, you know, Chinese companies have options and each company’s different, right? Because DeepSeek, for example, wants to have its hands on the hardware. So they don’t they’re they’re probably not gonna use anything overseas. They wanna have the actual hardware because that’s what that’s what they do. But Alibaba and ByteDance and companies, the hyperscalers that have data centers China. You know, they’re probably gonna they’re they have more options, right? And some of those H200s I think will probably go into could go into data centers outside China too. and then NVIDIA is selling the CPUs now are also really important for some of this. And so there’s no controls. It’s a weird loophole, but the Vera CPU, which is used with the Vera Rubin GPU architecture, can now be sold to China NVIDIA just this in the last couple of weeks is marketing that to China. That’s a very capable CPU, which could paired with other accelerators and used for AI training and other things, right? So that so the compute environment in China is really complicated by the because the are there, but they don’t cover everything. They don’t cover remote access. They don’t cover CPUs. and so Chinese companies now have some options here, but they’re, you know, but again, it’s like what is Here, right? It’s much more complicated for a Chinese AI developer than it is for OpenAI or Anthropic, and that’s that’s sort of theGrace Shao (51:10)Absolutely right. I’m really glad you brought up the H200s because I was gonna ask you about that. And it’s very interesting to learn that the CANN like system is trying to become more like CUDA and it makes sense if you’re trying to entice people to move over. Okay, I have a question, I don’t know how to ask it because I’ve heard it asked in both ways. Some people are saying, therefore, why is China still lagging behind if they’re capable of still getting access to certain ships and they’re so talented, right? Or the question could be asked in different kind of framing, which is why like wait, I just asked why are they so behind, right? Others are saying, why are they be able to catch up, play catch up if they’re so limited to, you know, generations ago. So, like, you know, the same question is basically being asked with different framing. At a high level, how do you view this right now? Because frankly, going back to your commentary even on open AI being able to just keep spending and keep purchasing. The most frontier GPUs, then the question is, is that price even justified if you can get almost frontier near frontier with, you know, four generations ago GPUs, then why do you need to spend so much on the latest, right? Like there’s a lot of discussion around that. Is the CapEx kind of justified? I guess this is a big question. See how you want to answer it. It’s complicated. Yeah noPaul Triolo (52:25)We should probably do a whole show just on that, because it’s complicated. Yeah, no, that’s a great question. And you’re you’re as you’re you’re really good at asking, you know, the really tough questions here. So I mean the and I think the d the difference then is that in the US, this idea of scaling, right? The scaling laws still hold. So the more GPUs you throw at training, the better the models will be. You know, that’s still sort of the view in the US. And it turns of that may be true to some degree, but there’s a lot of factors, for example, besides scaling that make models capable. There are these, there’s the harness thing, right? Which is the which is the ecosystem, the orchestration around the model. That’s really important in terms of the performance of the model. What tools can the model call, right? There’s a whole huge effort, you know, to standardize the calling of MCP protocol. which is used to connect the model to other applications. I just hooked up, for Claude. I gave Claude access to one of my brokerage accounts. And it can go in there and pull all the data on all my investments and analyze it, right? and so it does and it’s really good. And it learns, you know, more about you know certain other topics. A little bit well I have I had to sign away I had to tell the brokerage made me like you know sign away all the rights to any you know that happened.Grace Shao (53:38)That sounds so risky, Paul, and you’re so brave.Paul Triolo (53:50)So anyway, but the point is that the raw model and the scaling and the GPUs, it turns out that there’s more the scaling does still hold, right? I mean, Dario from MADA, where of course the CEO of Enthropic, you know, he was like discoverer of the scaling laws. And so you the major US labs like OpenAI and Enthropic and Google, and you know, there’s still the sense that the that the more GPUs throw at it and the more training you’re doing, you’re gonna get better models. But it turns out that like DeepSeek and others, there’s you know, through optimizations, because they don’t have access to unlimited compute, they figured out ways to optimize the and to enable them to run more cheaply. and that and that’s that’s affected the diffusion of the models. So if you’re when you talk about you know who’s ahead, there’s sort of raw model capability is one thing. And then there’s like who’s using the models, and everybody, it turns out that everybody doesn’t need the most advanced models. To run to run really useful applications, right? And so that’s where the Mabel fable and mythos thing, it turns out that you know people are now worried that the US government will, for example, cut off access to these models. And so why wouldn’t you use an open s a really capable open source model from China like GLM or Moonshot? Kimmy is very popular in the US. And you see in these recent reports that you know Coinbase and Microsoft and all these US companies are considering or using Chinese open source models in production, right? And so the question there is, you know, those models are exactly as good as the US models, but they’re pretty good, right? They’re good enough. And so it’s is the question of who’s winning the sort race, you know, is sort of less material in some sense because of these other factors, right? And so it turns out that the that both the model capability, once it’s near frontier, it’s good enough to run most things, right, that need. Some companies will still want to have the most cutting edge model. And also they’ll wanna have the issue of like their where their data is, right? They’ll wanna trust the company that’s that’s running their data whether it’s through an API. they’ll wanna trust that company their data. and maybe they don’t wanna they don’t wanna do that you Chinese model cloud, but they might be willing to run it on premises, a Chinese open source model and build on top of that. And that’s that’s that’s also what Fable and is sort of forced issue. Now companies are thinking, why do I want to give all my data to Anthropic or OpenAI when I can run a very good Chinese open source model on my own infrastructure and I can control the data and the and the security of that of that of my you know my business model. you may have seen Alex Carp’s sort of rant, people some people called it a ramp a couple days ago where he was talking about that issue where he was basically saying that, you know, don’t want to give all their data over to these model developers because then those model developers will compete with them for certain things. Which what’s happened like whatGrace Shao (56:41)Yeah, they will eat their lunch instead. And people also have a misconception that like when you use a Chinese model, it’s not like you’re giving the data to an open source Chinese model because you actually can self host that model I in your home countries. Anyway, I’m gonna start wrapping the conversation. I want to go big picture. Last okay. Last question on this. The dominant narrative, DC, is that, you know, AI policy circles, you know, circles often talk about, you know, whoever achieves AGI first, whatever that means these days. Will gain a decisive strategic advantage and ability to reshape global power. So it’s very, very scary. You know, how do you view this? Because through our conversation, what I’m hearing is that both sides are, you know, cautious, both sides are healthy and skeptical, but both are putting regulatory pressure, whether domestically on the companies or, you know, on protecting them from, I guess, foreign actors. Is this actually conducive for the future? Like how should we kind of view this? Because It also feels like from our conversation, a lot of these export controls, protective measures are not actually working or actually good for the industries domestically. So just a high level, like how do we understand this? Yeah.Paul Triolo (57:47)Yeah, great question. I think level, my concern, and I’ve written quite a about that, you know, if we race argument that the US is indeed to prolong the gap between US models and Chinese models so that when we reach something and I think AGI, I think we’re already at We’re already the models already passed the Turing test, right? But we’re talking about like artificial superintelligence where models are you know self they’re self-improving and they’re they’re coming up with really amazing new designs or weapons designs. If you look at the AI 2027 scenario, that’s sort of that’s sort of how people are thinking. Now I’m skeptical of that scenario because I think you know we’re still away from these models being to do you know the design super weapons and take over everything and so that one side who gets there first wins And then can kind of lord it over the other side. You know, this is essentially what Dario saying in his essays, like the machines of grace, and the adolescence of technology. He has said basically like AI, d democratic AI needs to win so that then that can be used to sort of force regime change in China, right? Or force authoritarians to sort of, you know, kawtow, if you will, to Western the Western governments. But I think that’s a That’s a I don’t like that framing because I think you know we’re not gonna wake up one morning and have that capability. It’s gonna be a gradual thing. and then the real issue is how governments deal with this? Like this whole issue of fable and mythos has forced the issue to the fore of how do governments deal with even just capabilities? This isn’t super intelligence, but this is like really good capability to detect vulnerabilities and software that exploited. And we don’t even have a framework for that, let alone for something more advanced intelligence. How would the government and industry work together on that, right? So there’s a couple things. One is, and I’ve written I just wrote a piece in Cairo Review about when does the government think about nationalizing the AI labs, right? Because people are using these analogies like these are nuclear weapons, right? Even though AI is not nuclear weapons. And I think it’s a very dangerous analogy. But people are saying, you know, this is a technology developed in the private sector, previously, weapon systems that were very capable were developed by government. And here we have a private sector ca capacity that’s that’s starting to edge towards weapons systems with cyber capabilities, for example. What is the gu how does the government do fit think about that? And the mythos thing, frankly, showed how unprepared the US government was for this, right? If you’re in the industry, you know that you knew that this was coming. I we talked about this last year at the Paris AI conference, right? You knew that this capability was coming, but nobody in the US government, the Trump administration was just saying innovation, innovation. China’s in the same way, right? How do you balance regulation and innovation? They want the companies to compete. So there’s no regulation. now Mythos and Fable have forced the issue of like, my God, well now we need to have some government role in determining, you know, how to test the models for certain capabilities and how to determine what’s a covered model and what should the conditions be around w how that model is released. And at least with Fable, we saw company, in this case Anthropic, have to, you know, put guardrails around the cyber capabilities of that model. And now they’ve finally the government has allowed them to release Fable. But that’s not there’s still a lot of questions around that, right? So the problem is if we’re in this, if we accept this race idea, then we’re never going to get collaboration between the US and China here, which I think is really dangerous because then, you know, malicious non state actors are gonna get access to this capability. and then, you know, the implications of that are really, really serious. And so the problem with the race idea is that it forces everything is it that then becomes distrust. The US distrusts China, China distrusts the US, you know, the US is gonna ban open could ban open source Chinese models. China could you know r restrict the release of open source models because they don’t want to contribute to The US developing capabilities. So we’re gonna get if we get into this race, then it’s a bad thing for everybody, I think. So we have the US China AI dialogue, which is coming up hopefully after the World AI conference in Shanghai, which I’ll be attending next week. and you know that’s gonna I think that’s the last chance. It’s the last chance for the US and Chinese governments to say, okay, we understand we don’t trust each other, but this is a threat, the threat of you know uncontrolled access to these models. It’s a exactly. It’s so it’s the last chance for the for governance to recognize that. And I think we I think that’s the mythos fable thing. The good news is that really drove I think this agreement in Beijing and Mart and May to between the t the two presidents to start talking about this. But it’s a complicated issue because ha you know, you got what are what is the goal here? How are you gonna agree on both sides to some limitations on this, right? And then how do you how does this translate into eventually a global agreement on putting guardrails around frontier AI models. It’s it’s it’s the problem is the technology is developing so fast that the ability of governments to keep up with this and come up with, you know, credible and viable structures to put some controls around this, you know, it’s really tough because there’s just not enough expertise in government. It’s going to probably have to be an independent, private sector led effort to do this. And this is This these are the kind things that are going to be discussed in week. I’m on a couple of panels, including some closed-door panels, that where these issues will be discussed. Now nothing’s gonna be decided next week in Shanghai, but I think the level of the discussion will be much higher because of fable and mythos and because the Chinese government is kind of freaked out about this. and you know, and the good news is that at least at some point the US and China will eventually sit down and try to begin this. Scott Bessent is gonna head up the US side and Vice Premier He I just did a piece with Alvin Graylin that you’ve probably seen in on the ASPI website, which tries to look at who’s gonna participate in this from both sides because there’s lots of lots of equities and we saw the that whole issue become complicated just in terms of deciding what to do about Fable. it was good in the sense that the governments to have a serious discussion of what are we going to do about this, right? So that’s the good news. But when I was at the way the finally, when I was at the World AI conference last year, I think it was Stuart Russell, my good friend Stuart Russell, who said, you know, he had talked leading CEO of a of a US lab, and he had said the best thing we can hope for in the next two years is a Chernobyl style event, right? Now think of what that means, right? This is the head of a lab. admitting that you know AI could lead to a very bad outcome here, right? and so this is where we are here in the summer of twenty six, where US and China both have leading labs. Governments don’t seem to know how to get a handle on what to do about that. But the US and China have to talk about it. Because if we get into this race, if we if we if we if we basically give up sort of say, okay, it’s gonna be a race, you know, and then th what will happen is something bad will happen and then and then people will say, my God, now we this. and you know, some people think that Fable Mythos thing is good because there were there was no bad event, you know, no loss of life or no nothing. But it did sort of force people to realize, okay, now we need to do something, you know, that’s a good thing. But still the political pressures and all these things we’ve been talking about, the export controls and everything, you know, there’s no there’s just no trust on either side. we’ve dug a deep hole in terms of and so Digging out of that, as you say, you know, to for the benefit of humanity is gonna be r a real challenge now. but you know, hopefully, as I say, in the next couple months, we’ll know better where that dialogue is gonna go and where China’s gonna be, for example, coming out of the World AI conference. They might announce the World AI Cooperation Organization, for example. I suspect they will announce that. Xi Jinping is coming, just to show you how important this issue is. Xi Jinping is coming to Shanghai. So the security arrangements around this conference are nuts. I’ve just been trying to figure out where I’m gonna be on different days. so that shows you how important it is. If Xi Jinping is coming, it’s important. and if Xi Jinping has agreed with President Trump to discuss this at some level, that’s that’s good. That’s good news. But as I say, I think this is like humanity’s last chance to get a handle on this because you know it’s the US and China have to agree. Everybody else matters, you know, there’s a big safety community, there’s other capable model developers in other places, but really the US and China are where ninety percent of the action is. and so if there is no agreement between the US and China, beginnings of an agreement around this, then you know, then all bets are off. And so I think this is a really the next couple months are really critical in this in this arena. And you know, the and that’s the technology continues advance and recursive self-improvement kick in. And as, you know, these things, you know, that’s not gonna stop. There have been all these efforts to say, hey, let’s let’s stop until we figure out what to do, right? Let’s pause, right? Who’s gonna pause at this point, right? I mean, those a year a year and a half ago that all these scientists, including it from China, signed on, like, we gotta pause, six month pause. The hard part is if you pause, you know, how do you decide when to start up again? Right? it’s it would be you know impossible. So nobody’s gonna agree to a pause. So therefore we need to agree on A minimally viable framework around governing these advanced models. And that’s what the goal is going to be in the next couple months. But you know, it’s it’s it’s really going to be hard because of the government the lag in capability in government you know so the mythos thing highlighted both need to do something but also wow like it’s the people who really understand these issues are still limited in numberGrace Shao (1:07:46)Yeah, we need more technical people in the government. Like actually every government. That’s the thing. Because this technology is not for the laymen to understand, frankly. Like you need someone who’s technical to understand it. But I think, okay, on that, like I’m feeling serious FOMO. I was planning on not going, but maybe I’ll go up. It seems like everyone is going. I was speaking to Alvin Graylin this morning actually. We’re working on a piece together. So it’s very interesting. I’m glad he’s he’s potentially going, you know, Ray Ma’s going, a bunch of people in the circles going. So You know what, like I think you’re right. Like I really do hope that something positive comes out of this. It’s just seems like it’s really hard to regulate something when regulation takes so much time. There’s so much bureaucracy that comes with it. And then on the other hand, like exactly to your point, AI doesn’t sleep. I was joking with my husband, I was like, I just want to summer. And he’s like, AI doesn’t summer, you can’t summer. And then he like being a tiger husband there. But you know, the reality is no one’s gonna stop right now, right? And likePaul Triolo (1:08:28)Right, right. I want three months off from all this, but catch up.Grace Shao (1:08:42)There’s a commercial interest and there’s also the com competition competitor like I guess even spirit in these researchers at this point. So it is gonna be incredibly hard. Yeah. So are these things gonna be nationalized? Do you think these like I mean, the irony and all this is like a look at my Cairo Review article? Yeah came out last month when I tried to layPaul Triolo (1:08:50)Right, and we also have massive IPOs coming up, right? We have anthropic that’s the other complicated Well, take a look at my Cairo Review article that just came out last month, and I tried to lay out how both the US and China view this. And I think y arguably already, you know, there is some there isn’t national you know, is gonna happen in different ways, but some level of nationalization is gonna have to happen, right? We’re already talking about open AI g you know, pr that the government taking a taking a investment or taking a share. OpenAI. So that’s kind of a there, right? so I you know it’ll be it’ll be different than nationalizing other industries, right? But yes, I think at some point it’s it’s hard to see the government leaving this capability in the hands of the private sector fully, right? because a important capability. And as we get closer and closer to more advanced you know AI and The idea of like loss of control, what happens, what happens if we lose control of the AI? all these things are out there. And so I think, but again, we can’t even figure out basic government role in, you know, how do we how do we determine what is a covered model and who is who is equipped to test that model, right? there are very there’s some efforts going on that I’m aware of to try to figure that out. Right. And it’s gonna but it’s gonna it’s not gonna be just the government. It’s gonna have to complicated, you know, body outside the government that’s that’s that’s that’s plugged into the government, kinda like the IEA, right, for n for nuclear for nuclear technology. But it’s right, there needs to be standards and there needs to be there needs to be a sort of neutral international body. But you know, we’re still quite a ways from that too. So we first have to get US and China to at least agree, Then building on that There could be the some new body. Now, again, I my the cynical view in the AI safety community I is that there has to be first a Chernobyl style event, right? That hopefully won’t be too serious before people get concentrated. Yeah, yeah, it does. It does. It’s ac absolutely Right. Right, right. But that’s sort of this that’s the worst case sort of cynical view within the AI safety community. But upcoming US China dialogue, it’s gonna be really important to see who’s participating, how serious it is, and you know, how quickly something can happen, right? Because the safety community has been arguing, we’re getting closer, we’re getting closer to artificial and it’s gonna take time. and so we need to start the serious discussion now. And that You know, that it started happening under the they were very serious about this, very thoughtful people. But then when Trump came in, it was basically let’s let her rip, right? Like US innovation is gonna dominate AI, and there was really a downplaying of governance. And then, you know, mythos sort of punctured that optimism in some sense and was like, okay, now we have to do But, you know, having not thought about that for a long time. You know, there were thoughtful people like Dean Ball and others who contributed to the AI Action Plan. And who’s now jumped up and out? Dean’s a great Dean’s great. I love Dean. He’s a great thinker on all these issues. So there are people out there who’ve been thinking about these, but it still turns out to be really, to, for example, set up a new organization. Late in the Biden administration, there was a discussion that Frontier AI is so different, right? It’s a different technology. You need a different regulatory structure around this. But it’s really hard to do that, to set up a whole new body and fund it and find but now I think people realize no, we this as another technology that we can just fit into our existing regulatory structure. It’s a different problem, it needs different capabilities, and so we need to rethink how to do this. I think that could happen too on both sides, both in China and the US, is okay, we need a we need to figure out a new structure here, an organization with the right authorities and the right capabilities and the right technical expertise to actually manage this problem, right? and I have I have a paper coming out with Alban’s not part of this paper, but I have a paper coming out with ASPE that’s looking at you know how one potential structure that could down this road having and both between both the US and China, right? A structure that includes the key players on both sides that would allow this to happen. But again, very tough you know, we’re it requires a sort of trust and concessions on both sides to figure out how to do this right. and, you know, the bilateral tensions that you see every day, right? are still a real impediment to this, right? Because AI in Washington, as you know, has become such a charged issue. You know, I mean we I mean people are talking about, you know,Grace Shao (1:13:59)But some of it’s talking point and some of it’s reality. I feel like at least in the business world, right? Sh maybe the policy world should have a little bit of that too. You know, what happens on the surface, what happens under understanding the reality and the realistic consequences that these talking points may lead to.Paul Triolo (1:14:01)Chinese. Right. Right, right. But Right. Well Right, no, that’s a great point. My the is that the during the later Trump first administration and the administration, this constituency developed around the AI issue, right? This sort of this weird sort of consensus that AI China, you know, we had to slow China down, we have to restrict these things. And that there’s a you know, that was in the in government, in the in the media, in think tanks. So there’s this huge sort of constituency of people. Who are wedded to the idea that we have to win over China at all costs on AI, right? and then there’s a group, the AI safety groups and others, and then people like me who were saying, well, no, that’s not the way that’s the sort of that zero sum thinking is gonna lead to disaster, right? and that we need to fig figure out a bet a better way. Like that the better way is how can we collaborate with China in these areas where we do respect national security concerns, but we don’t over index on them to the point where we can’t collaborate with China and then, you know, it’s a free for all and you know, bad things happen. and so that’s sort of where we are now. And I think the good thing is the Trump administration isn’t wedded necessarily totally to the previous administration’s approach to this, but it’s still it y you need smart people in D C like David Sachs and others who understand the industry and where the industry’s going and the technology who can kind of who are outsiders. outside the beltway who can actually look at this more holistically and say, okay, wow, we can we can work with China here, we can compete with them there, we can we can, you know, control certain things, but we need to figure out a way to skin this cat. We need to figure out a way to get to some basic level of agreement here. Otherwise, you know, we’re all in for a world of hurt, as that CEO of the of the lab admitted in a private setting I mean Chernobyl style events sounds pretty serious, right? And avoiding that needs to be something that focuses people, in DC and Beijing on, you know, how to how to howGrace Shao (1:16:23)Paul, I agree with you. You are full of knowledge and insights and you’re full of differentiated views, but there is one question I ask every single guest as we wrap up the conversation. what is one differentiated view you hold? You think that’s something just very against consensus?Paul Triolo (1:16:41)Well, I just think that technology controls and the idea of choke points really bad idea because we’re in a world as an interconnected world, right? And in fact, like when I’m working with clients across the AI stack every day. And when I look at US China, you know, y the degree of inner of interdependence and interconnectivity here is much deeper than people think, right? People are like, decoupling here and there. No. I mean if you really if you really look at the at what’s happening on a day-to-day basis and the complexity of supply chains, for example, the idea that we can simply decouple in AI or elsewhere is just, I mean, yes, we could do that, but the cost of that to the to the to the and to the companies and to global supply chains is just, you know, really fully accounted for that. So my I of like always coming back to the reality of okay, what is what is the what’s what’s what’s happening with businesses on the ground on a day to day basis and how are they being affected by this, right? And when you look at that level, you the sort of, you know, the thinking and comments that I hear that are just are very divorced from sort of that day-to-day reality of how interconnected the US and China have become over the last thirty years. And, you know, if we’re gonna indiscriminately, you know, pursue policies that where that collateral damage and the sort of the full cost benefit analysis done, you know, then it’s like, what’s you know, what we’re what are we doing here, so I’m always just I’m always just sort of arguing for a thinking about policy that’s based on a sort of a really deep understanding of the reality on the ground and you know how innovation happens and companies de-risk supply chains and how there are certain dependencies. For example, like rare earths turns out to be a real choke point, right? In the way that semiconductor technology is not, right? There’s just way around China’s Chinese company’s dominance of say samarium cobalt magnet production, right? That’s a real choke point. that will take ten years to you know to unravel. whereas other choke points that have been used on the US side are not really choke points. So I think they’re just my differentiation is the need to step back and look at this and think about like what is the point particular policy? Is that does it make sense? And is it is it having you know is the cost benefit sort Clearly on the side of, you know, too much cost and not enough benefit. and so that’s what I keep coming back to. And then we didn’t even talk about Taiwan. My also my sort of nobody few other people talk to is the impact of all this potentially on Taiwan and the risks around Taiwan. We work with companies every day and we do exercises, for example, about around a risk around Taiwan to their supply chains. You know, something short of a military exchange, which then there’s no de-risking. but you know, it turns out that, you know, Taiwan and supply chains and Asia in general are so intertwined that when you start pushing on some of these buttons, the worry I have the worry that you’re gonna, you know, you’re increase the potential for disaster there, you know, unintended or intended or whatever. and I think not enough people are thinking about that. they’re only thinking about you know, deterrence and arming Taiwan, think is frankly a sort of a mistaken way to problem. so anyway, so I think that out-of-the-box thinking and sort of getting a getting a getting away from the standard view which has developed over the last 30 years, you know, is necessary. And I just don’t see enough of that in Washington or Beijing. how do we rethink some of these things in the age of AI? So what we need a we need China policy for the age of AI and we need a technology policy for the age of AI, And I thinkGrace Shao (1:20:46)We need more dialogues. The thing is like I hear from so many people, whether they’re working on policy side or the actual researchers and developers, they’re like people aren’t talking officially because of all the geopolitical headwinds and noise. And then but actually, you know, obviously people talk, you know, behind the scenes, but we need more official dialogues to get to more fruitful results. I think more Yeah. So Paul, I’m glad you’re going to WAIC. You’re gonna be leading these dialogues.Paul Triolo (1:20:47)That’s where governments and we need more dialogue. Yeah. Mm-hmm. I’m with you, Grace. you. And I really appreciate your perspective. Well, I’m gonna try to contribute a little bit here and there. I mean, I love the people that I’ve met with a lot of the Chinese AI safety people, for example. They’re very thoughtful. they’re very good. and you know, in some areas ch China’s China’s Chinese you know, organizations and individuals are leading. But also there’ll be a lot of really good people from the broader AI safety community right? From the Future of Life Institute and from Concordia AI and s you know, really good Players who are really have smart people that are thinking about these problems also. So it’ll be a really good effort. Unfortunately, because it’s in China, you know, some of the leading US AI labs and some people and the US government, you know, will not be participating in this. It’s it’s seen as a sort of Chinese thing. but there will be the a the APEC meeting is happening just after this. And so I think there may be some there’ll be a US presence at the APEC meeting. And then as I said, you know, eventually. Probably shortly after this, I imagine that the US and China will kick off this AI dialogue. And so you’re right, dialogue is really critical here. And this is such a complicated issue that, you know, the sooner this dialogue gets kicked off and the sooner that they can they can, you know, feel each other both sides can feel each other out and get to the real issues. Yeah, exactly. Exactly.Grace Shao (1:22:36)Paul, I’ve taken up so much of your time today. I really appreciate it. I’ve learned so much. Can we please do this again sometime? I have more questions for you. You have you’re so knowledgeable, but thank you so much today. Thank you for your time.Paul Triolo (1:22:36)Yep. And thank you, Grace. I really appreciate your thoughtfulness and the and the thoughtfulness of your questions on these complicated issues. You really bring a lot to the conversation.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Future of mobility, a deep dive into the forces driving the Chinese EV revolution with Tu Le 17.06.2026 1ώ 22λ

  Hi all, I’m really scared to even share this episode because the last time I recorded an episode with Kyle Chan and mentioned cars, I got ripped online. So I just want to emphasize again that for car enthusiasts, I AM NOT A CAR person. I am here to learn. Haha, ok now that I’ve made that disclaimer…Joining me today is the ever-so-knowledgeable Tu Le. He is the founder and managing director of Sino Auto Insights, author of the SAI Weekly Substack, and co-host of the China EVs and More & At The Wheel podcasts.He has worked across Detroit, Silicon Valley, and China, so he views the industry from the inside, through the traditional auto industry, the tech industry, and the Chinese market.I wanted to do this episode almost as an educational primer, not just for you all but for myself as well. Most people now understand that Chinese EVs are competitive. But very few people understand why and how that is translating into the Physical AI space.We talked through the Chinese EV landscape, why traditional OEMs struggled to make good EVs, how autonomous driving fits in, how these carmakers are integrating AI, and why home appliance and smartphone companies like Huawei, Xiaomi, and Dreame are suddenly making cars.Follow Sino Auto Insights here: https://x.com/SinoAutoInsightFor consulting inquiries, go DM Tu Le on LinkedIn!Website: https://www.sinoautoinsights.com/Btw, I’m rebranding Differentiated Understanding to AI Proem Podcast.To find the previous episodes of Differentiated Understanding, see here.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.Season two will host a series of guests from early-stage investing, as well as builders, researchers, founders, and product managers. For more information on the podcast series, see here.Chapters00:00 Introduction to Tu Le and Sino Auto Insights04:28 Mapping the Chinese EV Industry09:19 The Rise of Xiaomi in the EV Market14:17 Understanding BYD’s Market Position17:54 Challenges for Traditional OEMs in EV Production31:29 The Role of Government Subsidies and Policies36:12 AI Integration in EVs and the Future of Mobility45:13 The Evolution of Brand Experience in EVs46:53 The Future of Manufacturing and Market Dynamics50:46 Safety Concerns in Rapid Development52:51 Current Landscape of Autonomous Driving in China57:51 Challenges in Deploying Autonomous Vehicles01:05:00 The Future of Mobility and Urban PlanningAI-generated transcript (for reference only)Grace Shao (00:00)Hi Tu. Thank you so much for joining us today. I’m really excited to have you on.Tu Le - Sino Auto Insights (00:04)Thanks for having me on, Grace.Grace Shao (00:06)Yeah, to start, why don’t you tell us a bit about yourself? We were just having this conversation right before recording. I find your background really fascinating.You know, you can talk to a very diverse group of kind of people. You run a successful consulting gig, a consulting company. Tell us about everything that you do.Tu Le - Sino Auto Insights (00:24)So name is Tu Le I’m the managing director at Sino Auto Insights. I also create content. I run or I co-host two podcasts, China EVs and more, and at the wheel with my co-hosts that are very, very good at what they do as well. And then I write a weekly newsletter, almost weekly anyways, called Sino Auto Insights Weekly that just kind of goes over my thoughts on what’s happening in the industry now globally every week. And I’m actually not Chinese. I’m Vietnamese. And I was born in Vietnam and moved to the United States when I was a year old and grew up right outside of Detroit. My whole family, youngest of eight, whole family’s automotive. So grew up car kid and did that for a few years before going back to grad school and moving to Silicon Valley to work for seven years. So that’s where the knowledge of the tech comes in, especially the hard tech, where hardware software integration is such an important part of creating a great user experience. And then I met a girl and in San Francisco. my girlfriend, who’s now my wife, was transferred by her company over to Beijing, where she was born. And I decided to pull the ripcord and and follow her over. And what we thought was going to be a three- or four-year assignment ended up being thirteen. And during this time I worked in automotive; I worked at a few Chinese EV e-commerce startups. And so that’s when I learned and experienced nine nine six myself for about two years. And yeah, it’s not fun, super intense, but again I wouldn’t trade those experiences for the world because it gives me the perspective that I have now. about eight years ago I saw this huge disconnect because EVs were becoming a thing because of Tesla. Companies like NIO and XPeng had just been founded. And you know, in Beijing, as you know, Grace, there’s a lot of the German OEMs, and so there was a bit of arrogance about how hard or how simple they thought software was and really, really being consumer focused as opposed to product focused. So I saw this opportunity, and I started this consultancy, Sino Auto Insights, and you know we’ve been growing since we’ve done traditional work. We’ve worked with the UK government, US government on things. And then also when I moved back four years ago from Beijing, I left during COVID. So August of 2022 and then November, December timeframe, China opens its border and says, What COVID? Come on in. So we didn’t know that was going to be the case. so we decided to move back. And we opened an office here in just outside of Detroit. And we’ve been helping more on the investment side, looking for investment opportunities, what’s around the corner, but also giving our clients a better understanding of the Chinese EV players and the battery players and what they’re doing outside of China. So it’s a very, very interesting time. The mobility space, as you know, Grace, involves now AI, silicon, data centers, data privacy, data security, batteries. So it’s just, just a tremendously unique sector that I get to be a part of.Grace Shao (03:55)Thank you so much for sharing your life story. First of all, kudos to your mother. Eight kids. Like, I don’t know how she did that. Like I have two and I’m already dying. And also, I love your personal touch, you know, why you moved to Beijing and just learning about your background. I think it’s super fascinating. You pointed one thing out. Like when I was living in Beijing in Shanghai as well, I met a lot of German OEM like employees and people kind of low key don’t know this, that there’s a huge German community i it in the huge like and and French as well. A lot of Europeans are actually working in China for these, especially like luxury vehicle companies. and a lot of them did relocate out of China during COVID times. And a lot of them I’ve even heard anecdotally from two friends who say they’re dying to get back because they were born in Munich or Frankfurt. just because they’re so bored.Tu Le - Sino Auto Insights (04:23)Huge. We hear those stories a lot, don’t we, Grace? We hear those stories a lot.Grace Shao (04:48)it’s just because it’s just the fast-paced energy in China. However, okay, COVID was crazy. China’s fast paced. Let’s get to that actual topic today. I wanna talk about EVs. I wanna learn everything from you. so before we get started, when we think of Chinese EVs, most people outside of China think of BYD. think of maybe like the few other ones you mentioned, like NIO X Peng. Now Xiaomi Dreame, which is crazy; essentially, these home appliance companies are going into the space as well. they are there are state-owned companies, there are old independent automakers, there are startups that we just talked about. And then some of them also produce batteries; some of them are, like I said, home appliance and phone companies. Basically, all of these different moving parts, they’re all coming into the same arena. Help us map out the industry first. Like to start with, who are the main players? What are the buckets? what does each group bring to the table or what’s their differentiating kind of offering? I know this is a very big question, but start with a big picture.Tu Le - Sino Auto Insights (05:45)So well, let me press rewind and kind of frame it and create more context as opposed to just we’ll we’ll zoom out and then we’ll zoom into the China market. So last year, twenty twenty-five, Toyota was the number one global automaker, eleven million units, around eleven, just over eleven million units. Volkswagen was number two at eight million. To give you a sense of scale, Tesla was one point six. million units and BYD was about 4.6, which makes them a top 10 automaker. The other top 10 automaker for the Chinese was Geely. Geely and everybody else outside of BYD and Tesla build ICEs and EVs. Or in China, they call them NEVs, new energy vehicles, which means that they’re battery electric vehicles plus plug-in hybrids, E Revs, and then fuel cells. So fuel cells, for our intents and purposes, are rounding error. So when we talk NEVs, we’re talking battery electric and plug-in hybrids and extended range electric vehicles. So Toyota’s been number one for a long, long time. And you know, the China market has been the number one passenger vehicle market since 2009, overtaking the United States. And now the China market is almost twice as big as the US market. If we add the European market, which is around 12 and a half, 13 million units, and the US market, which is around 15 and a half, 16 million units, it’s almost the same as China. And so the scale of the China market is enormous. And so to talk about EV specifically, I would create different sets of buckets. And I would look at BYD, Chery, Geely, Great Wall as separate companies, SAIC because they produce in the millions of units. Okay. And then this lower tier, I won’t say lower, but this other tier of EV makers, the NIO, the XPengs, the Li Autos, the Zekers, these are companies specifically that Western investors pay attention to because they’re traded publicly in the US. They’re in the hundreds of thousands of units. Go ahead. Yep.Grace Shao (07:58)I want to comment on this. So you’re actually separating them by the number of units versus their technology, because it seems like, just not like SAIC, they’re actually a traditional OEM company, but they’re also producing EVs, whatnot. So you’re actually categorized by number of sales versus, I guess, I don’t know, EV native like NIO and XPeng. Just help us understand what why that industry does that.Tu Le - Sino Auto Insights (08:18) so the automotive industry is very capital intensive. And so scale creates cost efficiencies. Okay, so if I buy 10 of something versus one of something, I’m gonna get a better price, generally speaking. And that’s why it’s so important that BYD has this enormous scale of 4.6 million units. And that’s through the traditional lens. Now It’s multi-layered as you’d mentioned. You know, they’re they’re EV only companies that we can talk about. But from the standpoint of scale and global reach, that’s where I’m really creating these separations because scale also creates flexibility because it’s gonna be harder for a company that only sells 300,000 units of anything to go global as opposed to someone that sells four point six million units of something. Because these companies, the BYDs, the Geelys, the Leap Motors, they all already ship and build and manufacture outside of China. Big, big steps. And so those are really kind of the uniqueness to some of those top-level guys that have the sales volume. They have the ability to go abroad. Because think of it just from a number standpoint, Grace, if we have capacity of half a million units. You and I run a car company. Building a hundred and fifty thousand unit factory is a huge consideration for us because we have to find demand somewhere for that hundred and fifty thousand units. Whereas if you have millions of units, 150,000 units isn’t that huge in a grand s in the grand scheme of things of sales and distribution. And so that’s where it’s a little bit easier for these larger companies to really command, you know, the pricing scale that they can negotiate over some of those smaller players. But you know, back to kind of how I would look at this. You know, the NIOs, the XPeng, the Leottos, they’re publicly traded in the US. That’s why there’s a lot of attention paid to them. But they’re still puppies. And you know, all of them shipped less than half a million units last year. Now Xiaomi is one of the newer players, but it is making a huge, huge impact because their automotive division, and you and I, I think, were there when Xiaomi first was founded in 2010 in Beijing. But their automotive division is less than six years old. And this year they’ll ship over half a million units. Contrast that with the NIO and XPeng, who’ve been around since 2014, 2015. They are barely shipping 500,000 units. So that tells you kind of the impact Xiaomi has had. Xiaomi only has two products. NIO and XPeng have four, five, six products. And so, now with Huawei, I would pull them into a very unique budget because they don’t build any cars. What they do is partner with other OEMs, whether they’re state-owned or non-state-owned, and offer the hardware and software stack. Okay. They call it HEMA, which is the Harmony in Mobility Alliance. And so there are companies like when you hear about Ito, Micstro, Stelato, Luxe; I won’t talk about who their partners are, but these are all using Huawei technology. And more and more, there are foreign companies that are using Huawei technology. So, Audi, I think Mercedes is using some of the Huawei technology in the China market. So we get into a lot of crossover when it comes to Huawei, because in order for Huawei to really, really legitimize their tech stack, they probably need more foreign automakers to sign up to it. And they’re really, really, really aggressive on trying to scale that part of the business because, you know, once they lost some of that handset business from North America. They try they’re trying to quickly find revenues to replace that. And I think focusing on the automotive sector was one of the ways they were thinking of doing that.Grace Shao (12:33) That’s super, super helpful just to get an understanding of the different buckets. so why is Xiaomi doing so well then actually? Because, you know, I saw a Xiaomi car, I think, maybe last year when I visited Beijing. It’s very sleek. It feels very nice, but just in comparison to Li Auto, XPeng, and NIO, which, as you said, they’ve been around for like more than a decade. They’re very, very they’re actually beautiful cars. I was quite like shocked when I first saw some of them during COVID time. Yeah, why is it that they’re just outbeating them? Is it because of the ecosystem? Is it because of what people are talking about with the control, the operating system? Because Xiaomi’s operating system is just significantly better because their software is better? Or what is it that’s driving consumers by them over others?Tu Le - Sino Auto Insights (13:20) So there’s a few things going on with Xiaomi. So the Su 7, which is their Sedan, if you squint, it kind of looks like a Porsche Cayenne. And then the U7, which is the yeah, so that helps, I think. And then the U7, which if you squint, looks a little bit like a Ferrari Purosangue, which is their Ferrari UV or FUV. So theyTu Le - Sino Auto Insights (13:46) Have borrowed design language from these two amazing automotive brands, and that’s helped them. But they have offered these vehicles at less than $40,000, starting at $40,000, with features that are very, very technology-forward. And one of the big reasons they’re successful is I bet. Grace, if I looked around your apartment, I would probably see a Xiaomi product, one or two at least. And in China, I didn’t know anyone who did not have some sort of Xiaomi air purifier, rice cooker, TV, computer, mobile phone. So the brand is ubiquitous in China. And that really helped Xiaomi when the vehicles launched, create this automatic instant demand because the brand is there: complete awareness of the brand, and there’s a lot of trust amongst Chinese consumers. And one of the important things about the China market versus the rest of the world is that anyone born after 1990 in China is a digital native. So they grew up with WeChat, Didi, Meituan, and Alibaba, you know, and Xiaomi. So, and these are all Chinese brands. So, they trust Chinese brands, not like their parents who only bought foreign brands. And I think that’s the larger shift in the Chinese Chinese market across sectors, consumers goods, you know, technology, high, you know, consumers products, now automotive. But also Xiaomi is very well connected across different, you know, product segments. They have an app that controls everything; it extends into the vehicle, and you would call that a consumer-focused product company. I think that resonates with a lot of Chinese consumers. So the impact that they’ve made is enormous. I haven’t even talked about how cool the cars are or what they can do. They’re breaking records at one of the most historic racetracks in Germany, the Nürburg Ring, and they’re beating the pants off of Porsche. So if I am Porsche and I know that Xiaomi is entering the European markets, Germany in 2027, I’m pretty worried. Because Porsche’s not doing well in the China market. And I think that’s the canary in the coal mine for a lot of companies in Europe, that Xiaomi’s gonna be a major player as long as they can continue to build these cool cars.Grace Shao (16:17) Super interesting. I would wanna look I wanna talk about Chinese vehicles, Chinese EV companies going global and going to Europe later. But to start, I wanna talk about the hype that BYD gets as well. Some of them cost less than even a hundred thousand RMB. You know, what is it about BYD, and how to understand them from a business perspective as well? Like they create their own batteries. They are getting into semis. They have their, you know, their whole whole integrated industrial platform. Like help us understand BYD.Tu Le - Sino Auto Insights (16:51) BYD’s story is amazing. In 2009, my first day in Beijing, there were BYDs. I got into a I I want to say a BYD cab, and it wasn’t great. It was not good. You could hear the exterior, the outside world, pretty, pretty clearly. And I I wasn’t feeling that safe in the vehicle, to be honest with you. But fast-forward to 2026, and Let’s just look at the last seven or eight years. Before COVID, BYD was shipping less than a million units. They’re at four point six in twenty twenty five. We’ll likely get to five million. They’re gonna be exporting about a million of those, a little over a million of those. So they’re currently in over 100 markets. So they are uber aggressive. And to your point, they’re very vertically integrated. BYD started out as a technology company that supplied batteries and other components to companies like Apple, and I want to say like Intel, but they got their experience and their scar tissue from working with some of the toughest technology companies. And that’s really kind of created s this resilience and ability to grow and scale and stay aggressive. Now they created a monster because companies like Geely and companies like Leap Motor are right on their heels, and they’re actually not doing that well. Their growth is flattening out, and the competition in China is super, super intense, hence the importance of them to export. Wang Chuang Fu is the founder CEO, and Stella Lee is the head of international. They’ve made it clear that they’re targeting Toyota to become the largest automaker in the world. And three years ago, they were going to be number one with a bullet, but now we’re seeing, as they scale and that denominator gets much bigger, double-digit growth is much harder to come by. Competition is catching up. They created the competition. They really, really were catalysts for all these other companies to build these sub $100,000 or sub-100 RB cars that are. Pretty amazing. Now, you make a great point because in China, BYD is the mass-market value car. Okay. But it creates an entry point for Chinese consumers that wouldn’t otherwise be able to purchase a vehicle, especially in big cities like Beijing and and Shanghai, where it’s, you know, getting a license plate is a challenge. And then finding parking and being able to pay for that is also very challenging, especially if you live right in the city center. And BYD has really, I think, in 10 years, 15 years, we’ll point to BYD as a democratizer of many, many, many things, not just mass market clean energy vehicles. And their focus on going international really, really put them behind a little bit on the technology curve. And Companies like XPeng are leaning into the technology. And Wang Chuang Fu has acknowledged that they’re a little bit behind on some of the features that other Chinese automakers are providing in the China market. But he’s determined, with his over a hundred thousand engineers, to really push that envelope to catch up and surpass some of their domestic competitors. Examples, recent examples of that. They’re launching a megawatt charger, or they’ve launched a megawatt charger in the China market. And that’s basically charging as fast as gas. You can charge fully within six, seven, eight minutes. And then also recently, just last week, I want to say, their intelligent driving system is called God’s Eye. And they are now providing a year’s worth of insurance. If there are any accidents while you’re using God’s Eye in China. So they’re putting their money where their mouth is. You don’t see Tesla doing that. And so they are willing to take that next step that everyone else begrudgingly has to follow them on. And, you know, one of the important things in, you know, I’d mentioned ten years ago that in 10 years we’ll say they’ve democratized things. The other thing is they’re offering God’s Eye as standard on many of their vehicles. And so think of it from the standpoint of not only in China, but to your point, now in Hong Kong, in Thailand, in Latin America, these people that have a 10,000, 12,000, 15,000 US dollar car, they might be able to drive themselves or at least have portions of the road where the vehicle has intelligent driving capabilities. And if BYD doesn’t do that. Intelligent driving doesn’t happen in those emerging markets for 10, 15 years at least. So they’re really, really pulling, I think begrudgingly, a lot of their competitors forward on that technology curve. And I applaud them for that. Now, at 4.6 million to get to 11 million, I think Wang Chuang Fu and Stella Lee appreciate more the level of management capability and the operational efficiency needed, like by a Toyota, to get to eleven million units. And so, but they’ve doubled down, and it sounds like they’re still determined to be a top two, top three player in the next five to seven years.Grace Shao (22:32) That’s really interesting that you pointed something out, which I didn’t notice at all. It’s the fact that they’re democratizing the technology, so it’ll be very interesting that they will actually introduce the kind of next-generation technology to these markets. So I guess bringing down their price right now is a long term strategy because once you capture that. you know, mind share, then they could always increase their prices later on.Tu Le - Sino Auto Insights (22:53)One of the most important things, Grace, is that there are two emotional buys in a person’s life generally, and that’s a house and a car. And in China, I think there’s around 60 hours, I wanna say, of research being done online, at the retailer, test driving, before you actually pull the trigger and buy something. And soGrace Shao (23:02) Yeah.Tu Le - Sino Auto Insights (23:17) If you and I- I don’t know if you have an iPhone or an Android phone, but if you lost your iPhone tomorrow, you’d be upset, but you’d walk into an Apple store and buy another one right away. But, you know, a car, people take consideration because it’s a reflection of who you are, who you want to be, you know, and really outside of your home and office, you spend most of your time in the car, especially if you live in Asia. And so That’s why it’s so important for them to be in these markets as a first mover. They create that awareness first, they build that trust first, and it helps them elbow out other players that might have equally impressive products, but because they came two or three years later, BYD already is in the mindset of a lot of these international consumers, especially in the emerging markets where a lot of times they’ll enter and six or seven months later, they’re the number one brand. In this segment, in that segment, or these segments that they enter.Grace Shao (24:19) Definitely. So let’s bring it back to traditional OEMs. You worked with some of them or you worked at some of them. Why did traditional automakers struggle to build compelling EVs now? Especially, you know, in China. The factories were there, the people were there. As you said in the beginning, you said some of them became a little bit complacent about the idea that, you know, Chinese consumers maybe just really liked luxury cars from Europe; you know, the default was buying Japanese cars for families. Why is it that almost every traditional OEM has produced a kind of crappy EV version?Tu Le - Sino Auto Insights (24:55) A lot of it has to do with, so let me first qualify this by saying if you ever talk to someone that tells you they predicted that the market was gonna move quickly over to clean energy vehicles in China, do not believe them because no one could have predicted how fast the market moved over in twenty twenty. We were in 2019, we’re at like one point two million units of NEVs relative to a twenty-two million unit base. Okay. And then in twenty twenty we were like one point three. Then we got to three point five, we got to six point five, we got to nine million, and last year we were close to eleven and a half, twelve, thirteen million units. And so we are currently over one of the inflection points, over 50 percent. So every one of every two cars is an NEV sold in China. Okay. And no one could have predicted that. So everyone was caught flat-footed if you’re a foreign automaker. Now let’s add in the fact that they’re analog companies, right? They’re not software companies. They’re not technology companies. And I bet today if you and I were to go into a boardroom or any meeting room in Detroit or Dearborn or Auburn Hills or Stuttgart, They’re still talking about the product. Okay. If we look at NIO, XPeng, Li Auto, their founders come from tech. Okay. They iterate. You know, they don’t, they, they ship product that’s good enough, and then they figure out what the bugs are, and then they create over there updates to fix those bugs. And whereas traditional automakers, they try to wring out as much profit as they can over a five-year period. And that’s because traditional product development cycles are a five-year period or a four-year period. The best company on the legacy side is Toyota, and they’re at around 30 months. Most good Chinese EV companies, like BYD or Zeekr, can go from clean sheet to job one. And job one means the first sellable vehicle off the production line. They can do that in about 15 months, which is absolutely insane. Now they don’t have the blinders on that say we can only do it a certain way. They’ve challenged everything all the way through. Simple things like where you might be you know a product engineer or component engineer and you you throw it over to me as a manufacturing engineer. Everything’s in serial. That’s why it takes so long sometimes. The Chinese EV makers, they do a lot of things in parallel; they simulate. A lot of safety tests, you know, validation to engineering validation tests, you know, manufacturing validation tests. They simulate a lot of that stuff, and it shrinks the timelines. And they treat manufacturing like a technology as opposed to, you know, an analog product. And if I’m being frank and honest, and I know that you want me to be that, the Volkswagens, the GMs, the Mercedes, the BM, they’re busy back in their home markets counting their money for a long time. You know, these narratives that they they steal IP. You know, is there IP theft? I think you and I would agree yes there is IP theft in China. But let’s qualify that: if there was IP theft in this instance, the Chinese automakers would make great ICE engines, right? Like gas engines, you would think. Exactly. So in this particular case, it it wasn’t because they stole this IP and it wasn’t because of subsidies. Because there have been subsidies since 2009, and it has been a substantial dollar figure, right? Tens of billions of dollars, if not hundreds. But if we look at Tesla, Tesla has not sold a vehicle.Tu Le - Sino Auto Insights (28:59) Globally without some form of subsidy. Okay. Full stop. They got their factory in Fremont for next to nothing. Shanghai Giga, tons of incentives by the local government. Li Chiang put that deal together. Guess what? He’s now the vice premier of China. So that was probably one of the reasons he got that promotion. So the idea that Chinese subsidies are bad, US subsidies are good, European subsidies are good. That needs to just stop. And at the end of the day, if you have those subsidies, it doesn’t automatically mean that you’re going to win. Because you probably remember this, Grace. X Peng Li Auto NIO in 2016, 2017, 2018, they were struggling. They’re almost bankrupt. And it wasn’t until December of 2019 that the first Model 3 rolled off the line in Shanghai Giga. That you really saw that hockey stick inflection point. Okay. So despite all the subsidies, despite all these promising EV startups, it took Tesla to really bring excitement to that market. And so you can credit Tesla for being the catalyst for EVs globally, not only in the United States, but China. I think we should acknowledge that they’re a huge part of why EVs became a thing in China. Now, because of the competition, all these competitors came in because of the subsidies, because of the the the RB that was being given out by local governments, you had a ton of players come in. And last year it got so competitive that the Chinese government was like, no mass, no mass, no involution. And so we have to acknowledge that the China market acr across a number of sectors, not just automotive, is likely the most brutal automotive market in the world. And if you’re able to survive out of this mess, you’re gonna be a very, very formidable competitor outside of the China market.Grace Shao (31:15) Definitely there’s the Musk effect, I think, on you know, Tesla bringing out EVs, and now we’re seeing that with humanoids again. Cause with Elon Musk obsessing with humanoids, we’re seeing the world obsessing with humanoids. We’ll talk a bit about that later. I appreciate you giving kind of the backdrop of the history of China’s EV space. Tell us about the industrial policy push though, because you mentioned subsidies as a very vague term. But what kind of subsidies or industrial push do you think China actually gave this industry to bolster it?Tu Le - Sino Auto Insights (31:48) The Chinese government has the ability to long-term plan. That doesn’t happen in the United States. Unfortunately, and it’s really costing us in a lot of areas and sectors. And we’re seeing these sectors that would normally be pretty strong be a pretty competitive struggle against the Chinese. But l let’s say in 2009, the Chinese government looked at manufacturing as a pillar industry. That supported jobs. And within the manufacturing sector, they wanted to become number one in batteries, number one in silicon. And silicon fabrication. Not only silicon design, but silicon fabrication, automotive EV manufacturing, and then, you know, and that all supports this notion that they’re the world’s factory. Okay. And they doubled down on that. Now The types of subsidies, whether it’s tax abatements for land, whether it’s discounts on building factories, whether it’s purchase subsidies or you know, tax abatements for the consumer who don’t have to pay taxes on buying EVs. To your point, in Hong Kong, same thing, right? Similar things going on. I I wanna I wanna stress that anywhere in the world, emerging technologies, in order for them to become really ubiquitous and blossom, governments need to put their thumbs on the scale. Okay. So it’s not just a Chinese thing. The United States was ready to give consumers $7,500 for every car, every EV that they bought. Okay. So it’s not just a China thing. Norway, which has a ton of oil money, used that oil money to subsidize EVs. And now the take rate in Norway is over 90%. Now their market is tiny, 400, 500,000, 600,000 units a year, but nonetheless, it’s another example of the government putting the thumb on the scale. In Germany and other parts of Europe, there are also subsidies for EV purchases. And that’s also when you saw growth rates a little bit higher than they are now because they’ve taken away a lot of those subsidies. So again, emerging technologies need help. And normally the governments in any one of these countries need to step in. The exception is kind of the UK, which has done pretty well on EV adoption, the growth of EV adoption, despite not having a ton of subsidies on the consumer side. But that being said, it these subsidies, this focus, this diligence, this perseverance, this investment over time. It didn’t look like, let’s say, 2010, 2012, 2014; it didn’t look like it was going to really, really work out that well. And then all of a sudden COVID happens, and you get this huge spike in demand for electric vehicles or clean energy vehicles or new energy vehicles. Now, my quick story. You know, it’s it’s aGrace Shao (34:51) Why is that?It’s like we’re locked in our homes. So why do we need EVs?Tu Le - Sino Auto Insights (34:58) It’s a weird phenomenon and and I haven’t read anything or talked to anyone that can I you know, I th we can try to theorize. For me, it’s like, yes, we’re locked in. I wouldn’t say we’re locked in our homes, but we’re locked in the country, and we don’t travel, so maybe, you know, we spend money on something that we think might make us happy or something like that, right? Like, I can’t tell you why outside of, okay, Tesla starts building in 2020. And and and a quick story about my COVID experience. My family went back to Michigan. So I was sending my kids to a local school. So they had basically the month of January off in 2020. We were going to go back to the United States for three weeks to visit family. And the return flight was canceled. After two weeks in the US because China was starting to have COVID. And we bought two two one-way plane tickets that were canceled. And then a third one where we finally got to fly into Narita, stay there a night, fly to Shanghai, and then to Beijing, and then quarantine for two weeks. And we didn’t leave China for two and a half years at that point. The day we got back, the day after we got back, the Chinese government closed the border. And during that time, Grace, you saw more and more and more of these green plates. And I I was just amazed because think of all of the things that need to happen for demand and supply to work together. Because it’s not just, okay, we can produce these things, no problem. There needs to be charging infrastructure; batteries need to be scaled up to support, you know, the EVs. And then creating this awareness, creating this excitement. The NIOs and XPengs and Li Autos weren’t able to do it on their own. The BYDs weren’t able to do it on their own. But on the backs of the the the Chinese consumers really, really trusting the Tesla brand. And we’ve seen that the Tesla brand in the China market is extremely resilient because they’re still selling quite a few, despite not upgrading their vehicle or updating their vehicle in a number of years. And so it is a strange phenomenon that I can’t definitively tell you why. I just know I saw more and more green plates when I was in Beijing and Shanghai over twenty-twenty through twenty-twenty-four or twenty-twenty-two, so.Grace Shao (37:41) Yeah, it’s an interesting phenomenon. And like every single tech founder or AI founder I’ve ever met basically says they drive a Tesla. So contrary to what you were saying earlier, you’re like, a lot of Chinese younger generation actually prefer the Chinese consumer brands. There’s something about Elon Musk and his brand in China. People like love it, worship it, wanna become him, whatever. So all of all the founders and tech people still drive Tesla. I I wanna bring it to the next point, which is on AI and tech, actually. So we talked about EVs, and I know I can ask you like 3,000 more questions on this, but I wanna understand the AI element to all of this now, because essentially EVs are not like old cars; they are tech-first cars, right? Like you said. But because they're tech-first cars, they seem to have integrated. AI or hardware plus software much more seamlessly. We’re seeing like Li Xiang push out like even wearables like glasses that can control their cars. we all know all the mentioned brands just now have voice control. They all have some sort of AI already embedded in them. Now it’s just like giving them a more formalized name. Autonomous driving obviously is a form of AI. We can talk about that as well, but help us understand: for people, for these cars, are these kind of like software-hardware integrated really that seamlessly, or are they struggling as well? And then on top of that, you’ve said something I think in one of the podcasts before is that you push back on the phrase software-defined vehicle. Tu Le - Sino Auto Insights (39:22) So the terms mobile phone on wheels and software-defined vehicles, to me, and I would love your opinion on this, Grace. That tells me that these people don’t know technology or understand technology or have worked in the technology space because software doesn’t define anything. Software, AI, you know, silicon, these are all tools that create a compelling user experience. Now, you put them together, you design them well, you combine them with the right hardware and the software that instructs the hardware what to do, when to do it, how fast to do it, that creates the user experience. And I’m an Apple alum, so I learned that early on. It was really, really drilled in my head that the user experience, the stickiness, that creates the brand, that creates the brand loyalty. Okay. And what the Chinese automakers are doing right now is like throwing spaghetti on the wall to see what sticks. And because of the enormous pressure from competition, they just try to be first. Okay. What they likely need to do, and I don’t know if they’re going to be able to do this in the next 18, 24 months, just because I don’t see competition really, really slowing down in the China market. Is to take a step back, you know, I had a conversation with Sam Livingstone and Matt Mechelvoigue talking about the Ferrari Luce and some of the missteps. And part of that is, what does this mean to the NIO brand? What does this mean to the XPeng brand? Because that is what creates the awareness, the trust, and the loyalty; like everything just kind of makes sense. Simple design is really, really, really hard. Johnny I’ve has said that many, many times. In I’m sure. I would again love your opinion because I’m a Westerner, so I’m not used to Chinese apps that have a million things popping up at me or Chinese websites that have all these windows and all these lights blinking and stuff like that. It’s it’s a little intimidating to me. And I feel like front consoles of Chinese EVs are still a little bit overwhelming. Now, if we look at Xiaomi, I think they do it pretty well. They could They can improve, but they have years and years of experience among their teams to build out consumer experiences. Okay. And to me, like the reason I don’t like mobile phone on wheels is because a mobile phone can’t run you over and kill you. Okay. So anything automotive grade is a completely next level thing. So a big thing on the battery side is energy storage systems. Okay. A battery for an energy storage system is not automotive grade. It needs to be bulletproof if it’s automotive grade. So the level of engineering and manufacturing quality needs to be much, much higher. That’s why when we oversimplify it like that, I don’t think people appreciate the amount of effort and level of detail that needs to be had for putting something on automotive grade. And then software-defined again, it just tells me that these guys. think software is the end-all, be-all. Software, if software, AI, and hardware is running great, you don’t notice it. You just experience, you just have a great experience. If I have to say this hardware is not working or that hardware is not hard, then the automotive designers and engineers need to go back to the drawing board because something’s wrong. and and that’s what I want to emphasize being an Apple alum, that that hardware and software integration is is is what is going to create and differentiate you in the market long term. And one thing I will point out about Apple is that because they have a closed system and they don’t have to Frankenstein a bunch of disparate, you know, firmware together because they control the design of the hardware and the software and integrating, well. They might integrate third-party AI now because they’re super behind. But that’s kind of the only third-party thing that they’re doing. but eventually I’m sure they’re gonna look to to create a native AI support system for their ecosystem. But I I think that’s the huge differentiator. Now, is it realistic for an automaker to have a closed system and control so much? If you’re Tesla, maybe, but you started. on day one as a closed system. Okay. It’s gonna be extremely difficult, extremely, extremely difficult for most other automakers to do this. Now, you know, one of the areas I think you wanted to talk about was partnerships. And this is where the Chinese automakers are getting a lot of credit through the announcements of all these partnerships with their Western counterparts. And the important thing is that there’ve always been partnerships in the China market. You know, it’s been a requirement of the Chinese government historically, you know, with the exception of the the the Tesla factory in Shanghai recently. But now these partnerships are bleeding into Europe, they’re bleeding into North America, and we’ll continue to see that as long as the Chinese are pushing the envelope on innovation. But again, Grace, can the legacy automakers use somebody else’s tools? Again, they’re tools. Can they use somebody else’s tools? To create a Volkswagen experience, a Volkswagen brand experience that you know historically is this way or that way. Okay, like you trust Volvo, right? You don’t like their new cars, but can if Volvo is using Geely software, Geely AI, Geely hardware, or you know, like Geely qualified hardware, is it still gonna feel like a Volvo to you? And I think those are kind of the important things moving forward. that’s gonna differentiate some of the legacy automakers and some of the better Chinese EV makers from the rest of the field.Grace Shao (45:28) I feel like listening to you explain this actually makes me feel like there’s gonna be two fragments of the market where the hardware people, like the people who still want the best hardware experience, will still go with a traditional OEM because you will still get the best craftsmanship, get the best kind of hardware experience, right? But if you are gonna go for an AI-native experience as we go forward with this, you know, you want the best voice control, you want the best. Whatever interaction with your AI agent within your car, then it would make, as you said, it would be extremely hard for a Volvo to use someone else’s software. So wouldn’t a company with their own software actually have the advantage of building that? So like a Xiaomi or Tesla, right? Like your point, you have your closed ecosystem, you have your existing s software, you have everything you need to make the experience better. But that said, the car might not be as sleek as an Audi, Mercedes, whatnot, right? I don’t. What do you think?Tu Le - Sino Auto Insights (46:32) Well, I think that when you get to clean energy vehicles, so especially battery electric vehicles, manufacturing is is simplified by orders of magnitude. So the GMs, the Volkswagens, the Porsche’s, they all Mercedes, they all have entire powertrain divisions that only work on the engine. Imagine those entire departments effectively going away. Okay. Electric motors I’m oversimplifying this, but they’re fairly commoditized. Okay. They’re super fast, super efficient, generally speaking. Automotive grade is something different than everything else again. But that is really going to be the differentiator moving forward. Grace, if I told you that in 15 to 17 years, maybe less than that, building a car is going to be commoditized. So there’s no value in that aspect or part of it. Now, with the world being as bifurcated as it is, especially in North America, that doesn’t want Chinese battery cells in North American vehicles for now. Maybe it’s a longer timeline, but effectively China is really, really creating or forcing other companies to rethink how they manufacture things, how they develop things in the spaghetti on the wall. There will be some spaghetti that sticks for the Chinese automakers. And you better believe that that copycatting is gonna be reversed now. The Europeans and the North American car companies are really, really going to create their own versions of X, Y, and Z. Now, can we say that it was innovated and perfected in the China market? Probably moving forward in the next three, five, seven years, but we also need to look at the demo. Right? Because you had mentioned some people want performance, some people want the digital experience. And I think a lot of that is gonna be is gonna correlate to what the demographic is, because a BMW or Mercedes owner in China is around twenty, twenty-five years old, younger than in Europe and North America. As a you know, as an old man, I have different needs than you do, as a as a young woman. So What I like in a car is going to be different than what you like, than what your husband likes, than what my wife likes. And I think that’s where the Chinese are gonna have to play the global game. Okay. Now they need a solid foundation of extremely high sales in their domestic market to create the flexibility to sell abroad and to sell at a premium abroad. But are all these digital features And technological advancements going to resonate with a 60 year old year old European man in Germany who is used to driving a BMW? Probably not. Okay. But one of the other big advantages is that the Chinese have is, and I’ll give you a quick example. Friend Nick Carey, who writes for Reuters, last year he wrote an article about Chery taking six weeks to change the suspension and steering system in an Omata 5 because they were shipping the China-spec version of the Omata 5 to Europe and the Europeans were like, yeah, this steering is way too mushy and the feel isn’t there. The Europeans will not like this. And so over six weeks they qualified new parts, they updated the software through OTAs and firmware and then shipped the new product And that would take a year in in at least a year at most legacy automakers because of the layers of bureaucracy, the approvals needed. But this was done in six weeks. Now that’s also a reflection of the nine-six in China.Grace Shao (50:24) Does that not frighten you a little though? Because like how fast, like you mentioned, how fast they ship, w I wanna ask something slightly sensitive. Then what about the safety of these cars, right?Tu Le - Sino Auto Insights (50:36) Well, if you asked the Chinese automakers, they will assure you that they’re not cutting corners. Okay. And I have driven many of these Chinese cars. Now I d I haven’t owned one for 10 years. So long-term efficiency and safety, I don’t know. Most people don’t, because a lot of these cars have been on the road for less than five years. But you know, when you talk to the automakers, and again, a lot of these people come from the automotive space. A lot of the leadership of some of these Chinese companies, they come from the Mercedes, the Volkswagen groups. And so they do have some visibility into how things are being done differently. But I would also counter what you just said with yes, there’s a little bit of risk with cutting so quickly the product development and so severely the product development cycle, but Also, how things traditionally work at large conglomerates and in even governments is there’s something new that’s happening. We’ll add a layer. There’s something new happening; we’ll add a layer. Technology changes, we’ll add a layer. So no one ever takes a step back and says, These 15 layers, does this still make sense? Because I mean, that’s kind of the definition of bureaucracy, right? So time will tell. I do I feel not safe in these cars? No. you know, and I’ve driven dozens of them, soGrace Shao (52:01) No, I was playing devil’s advocate. LikeI’ve been in so many of these in China, and you know, especially across Asia. But I just think the it’s just people tend to ask questions like, it’s so short then. If you’re shipping them out within a year, what kind of corners are you cutting? And then thus the question is easy to say. The next question is, is it safe? Right. But I think I I I kind of feel you on the point. If it’s completely new, we treat it like a startup; it’s innovative. There’s a lack of bureaucracy, there’s a lack of this is how we do things. Then you actually can just get things done much faster. I’m mindful of time. I want to ask you some questions beyond the traditional car makers and whatnot. Help us understand where China is with autonomous driving right now. Who are the main players and just roughly understand, you know, who are the ones kind of competing with Waymo, who’s Pony AI, right? Like who’s We Ride? I know again, it’s a super big question, but let me just throw this to you like open-ended.Tu Le - Sino Auto Insights (52:57) Let me kind of close out that last topic that we were talking about with food for thought in something that I know you know as well, but maybe your audience isn’t quite aware of. If any Chinese company is found to be cutting corners in the China market, the Chinese government would not look kindly on that. And there would be severe consequences, right? So I think there’s this healthy fear of If we are cutting corners and we’re found out, we’re gonna be in a lot of trouble. There’s not gonna be years of litigation, there’s gonna be severe penalties right away. And I think that healthy fear motivates many of these Chinese automaker leaders to stay on the right path. Right now, again, time will tell, but to pivot towards your question about autonomous vehicles, so Waymo is the global standard. I think most people would acknowledge that. They’re in many, many markets. They’re entering foreign markets. But there to your point, there is WeRide, there’s Pony, and then there’s Baidu. These are the three largest players in China currently. But then in Apollo Go, yep. And and and unfortunately, when I was in Beijing last month.Grace Shao (54:13) I do is call Apollo, right? Or yeah.Tu Le - Sino Auto Insights (54:22) Or two months ago, Apollo Go was not running because in Wuhan about a thousand of them or a hundred of them were on the roads and they just turned into bricks, right? On the roads. And so they had stopped the pilot programs. I don’t know if they’re running again, but I normally when I’m back in China will try out all these systems for the latest software to make sure to just kind of see, feel, understand, what’s going on and and what’s unique about China is that the feel is a little bit different in each of these cities just because how people drive is so different in in different cities. Grace Shao (55:02) This is something I feel like no one understand if you don’t live if you haven’t lived in China. Like people in Beijing are just aggressively wild. Like people don’t realize this. Tu Le - Sino Auto Insights (55:11) What? Aggressive? man. So I’ve driven in like Changsha, I’ve driven in these tier two cities, and you’re like,Grace Shao (55:17) Okay, I haven’t driven into your two cities. I just haha I usually just get a car there. Like, I I I already think Beijing is so terrifying. Like, I I start driving 16 years old in, and I refuse to drive in Beijing, and because you stop the car for someone to pass. Next thing you know, like 10 cars have passed, like 20 bikes passed you, 30 pedestrians, and you’re still there, and then there’s like 20 people behind you honking you, and you’re just like, yeah.Tu Le - Sino Auto Insights (55:33) my goodness. Yeah, so a fun, quick funny story. My wife used to be very worried because I would get super fired up when I’m driving in China. For some reason, I learned to compartmentalize it because I would get super upset. And then when I got out of the car, I would just not be upset. I don’t know how I did it, but because my wife didn’t want me to take yeah, I it was just.Grace Shao (56:04) Like you have to. Because you’re like constantly road raging. Anyway.Tu Le - Sino Auto Insights (56:10) And so what you’re talking about is called cutting in. And if you have a meter of space between you and the car ahead of you, someone will cut in. Someone will cut in for sure. And if you’re not used to that, someone will cut in, and then there will be a San Luncha delivery vehicle turning the opposite way. And your head needs to be on a swivel. And it is quite an experience. It’s similar, and I won’t say similar, but it has a similar feel as Southeast Asia because it’s so crazy. And it just kind of works in Southeast Asia. And it doesn’t work super well in China because there’s traffic jams all over the place. But Pony and WeRide are trying to help some of that stuff. Let me segue to that.Grace Shao (56:57) Yeah, so how does it work though? Like that’s my point. Like how do these autonomous driving cars work if, you know, people are so unpredictable? The whole idea is they’re supposed to predict what the car is gonna do, but you know, they just like zigzag and people just pop out of nowhere. Like i is it safe? Like what’s really a holding up, like what’s a bottleneck of deploying these at scale right now? Is it regulation, is technology, or is it just the craziness of the roads?Tu Le - Sino Auto Insights (57:28) Again, let’s do a 30,000-foot level. We ride, pony. They’re both publicly traded in the West. And so I think there are Western investors that know who they are. They’re not as large as Baidu Apollo Go, which has, I want to say, well over a few thousand cars on the road in China pilot programs in a dozen cities. But Pony and WeRide are also moving aggressively outside of China, partnering with Uber, partnering with other companies, ride-hailing companies, and there’s pressure because they’re publicly traded to really scale and create some profitability. Whereas Waymo is still owned by Alphabet. So I think they have pr internal pressure, but not pressure from external markets. in the difference between China and the US, because at the end of the day, these are really the only two players that have multiple horses in this race. Now in the UK, there’s a company called Wave. And I think there would be other European players that would argue that, hey, we’re also a major player, but let’s, for the intents and purposes, oversimplify this by saying there’s the US and the Chinese players. In China, there is this first wave of A V companies: the Werides, the Pony AIs, the Baidus, and then there’s this other wave, and we’re only talking about robotaxis. Because on the commercial trucking side, on the slow-moving delivery vehicles, we also have autonomous vehicle players. But for robotaxis, there’s this second wave. And they’re more of an asset-light company, autonomous vehicle startup. So like DeepRoute, Momenta, QCraft, you know, they’re what they’re doing is partnering with traditional OEMs in China to get their stacks onto these vehicles. DeepRoute, for instance, is working directly with Great Wall to integrate their hardware and software stack into the design of the vehicle. So even before, because what we normally see right now, Grace, is a car with lidar, sonar, radar bolted on as, you know, an afterthought. But once these companies are working with the traditional OEMs, they can design them, and it looks like part of the form of the vehicle as opposed to like this bolt-on after the fact. And the convergence between RoboTaxi Company and traditional OEM is blurring. And I’d mentioned earlier that Huawei also has a significant stack. So they’re a player as well. But Huawei, as far as I know, is not getting into the RoboTaxi space. But they’re going to move into level three intelligent driving, and they’re hoping to be in millions of vehicles in China within the next few years. And that’s where it’s very different in China because there’s not a lot of convergence going on in the US market. Now we know about Neuro, we know about Zoox, we know about Waymo, and with the exception of Neuro, who’s working with Lucid to put their stack on the Gravity for a premium experience. Most of these companies are not working with OEMs. And then the OEMs have their own systems as well. And that’s the big difference between the China market and the US market. And what we’ll likely see is a bifurcation of the US market being primarily North American autonomous vehicle providers working with the Ubers and the Lyfts to create that larger install base to try to reach a broader audience. And That’s one of the big reasons why these companies are working with the ride-hailing companies, because it’s gonna be hard for Pony to attract 100 million users. Whereas if I partner with Uber, my install base is 160 million global users. And so that creates an opportunity. And I think long term, Uber sees Robotaxis and Evital as their profit drivers. And you know, the delivery services and all these ancillary mobility services as a way to increase their install base but not make a ton of money. And and soGrace Shao (1:01:45) Wouldn’t that cannibalize our own business, existing business a little bit?Tu Le - Sino Auto Insights (1:01:51) Yeah, and you know that’s that’s the that’s that’s the million-dollar question because Uber is now also buying its own autonomous vehicles. So not only is it a ride hailing platform, but it’s a fleet manager now. And that changes the economics of their yeah, exactly. So that really changes the economics on their balance sheet. Okay. So I I don’tGrace Shao (1:02:06) hedging.Tu Le - Sino Auto Insights (1:02:17) To your point, I do think they are kind of hedging their bets a little bit. Because if to answer your question, we should separate autonomous vehicles into those that use lidar and those do not use lidar. Tesla does not use lidar. Wave, which is the UK company based out of London, does not use LIDAR now. We can have, and I’m sure you’re well- I think you’re talking to some expert in a couple of weeks, so maybe you can ask them to use lidar or not to use lidar. they are using lidar, so I’m sure he’ll he’ll say that lidar is necessary, but it’s more philosophical now, right? It’s more philosophical because Tesla can’t all of a sudden put lidar because it changes their whole system. Okay. But to me, LIDAR.Tu Le - Sino Auto Insights (1:03:06) Prices have gone down so significantly that creating another redundancy and, you know, kind of creating that sensor fusion with multiple sensors and lidar creates a safer environment, I would think. but again, it it it it’s an interesting thing that I don’t think a lot of people definitively can answer. I’m sorry?Grace Shao (1:03:29)It’s a philosophical choice. Or is it actually a design choice because the vehicle already can- like, you cannot put a LIDAR on top of it because Teslas cannot use LiDAR. Like, is there a reason?Tu Le - Sino Auto Insights (1:03:42) So, to me, it was an engineering choice at first, but now it’s more philosophical because now you’d need to change your system if you all of a sudden incorporate lidar into it. Right. And for Elon, I think it’s also a mienza thing because he’s been so strong against LIDAR that if he turns it around, now don’t get me wrong, he says things sometimes thatGrace Shao (1:04:03) Yeah.Tu Le - Sino Auto Insights (1:04:09) Never come true or haven’t come true yet. So that’s kind of the crazy thing. But I think LiDAR, that’s one of those things where I think he’s willing to die on that that doesn’t need lidar. But anyways, I don’t want to get into this this this discussion about lidar, but but but but to the bifurcation thing. The reason I say bifurcation is becauseGrace Shao (1:04:24) Okay. It’s gets getting too technical. All right.Tu Le - Sino Auto Insights (1:04:34) We know that Europe has a strong data security, data sovereignty policy, security, privacy, and sovereignty. So will Europe allow the Chinese autonomous vehicle makers to ship European data back to China to the servers to train the models? And/or so the United States, because we have a ton of allies, China has a ton of allies, that’s why. It’s likely that the Chinese allies would use the Chinese systems first, and then the US allies would incorporate the Waymos and the Zoox. That’s kind of, and I’m oversimplifying this, but because the Trump administration has kind of poked the eye of a lot of our traditional allies. So maybe they wouldn’t want our autonomous vehicles on their roads. But I’ve ridden in all of these systems.Grace Shao (1:05:08) I see.Yeah, like Canada might be saying no these days. Sorry, I’m just going. It’s like really late for me and I’m just thinking about yes, but like now Canada’s gonna have eight Chinese EVs. Now American cars aren’t gonna sell there now; American Waymo’s not gonna be in Canada.Tu Le - Sino Auto Insights (1:05:28) Yeah, yeah. No, you’re well, we could do an entire episode about all of that stuff just between North America, right? So I think that would be an interesting conversation. But you know, that’s exactly my point, right? Like, traditionally, prior to Trump, I think Waymo was going to rely on our allies to launch its services. And if we look at the Middle East and India, theyTu Le - Sino Auto Insights (1:06:06) kind of want to be Switzerland a little bit. So because they they they they don’t want to take one side over another. India’s the same thing. And these are potential markets where both of them will compete. Famously, in London by the end of this year, early next year, Waymo and Do will all be testing and rolling out pilots. So next time you’re in London, if it’s early next year, you might be able to try all three systems in the same place on the same streets, which I think is gonna be a very, very, very unique experience. because I don’t think there’s gonna be many cities that you’re gonna be able to do that in over the next three, four, five years. And, you know, at the end of the day, is it a data thing? Is itGrace Shao (1:06:47) That’s pretty crazy.Tu Le - Sino Auto Insights (1:07:02)Who has the most data? Who has the most robust edge case data? Is that ultimately who wins? Or does AI really change the game? Because if it’s about data, if it’s about kind of real-world miles, then you would think the Toyotas and the Volkswagens would have a distinct advantage because guess what? They put 11 million cars a year on the road. Okay. So a company like a deep route, a company like a wave would love to work with these companies that high have high sales volume. But what is the great equalizer? If you talk to Elon, it’s his system because again, everything is closed, everything is native to the Tesla system. FSD is the best intelligent driving system. Is it better than Waymo? And will there ultimately be a convergence between level three and then level four? You know, can a Waymo system compete directly versus a Tesla because and and I don’t have answers to that. And that’s what makes the industry so interesting because the politics of things change, the technology changes, and then the commercialization opportunities change as well. And what I do know is that Waymo is backed by one of the most valuable companies in the world, which means that they haveGrace Shao (1:08:09) Mm-hmm.Tu Le - Sino Auto Insights (1:08:28)A huge check to help them get to scaling. And one th other differentiator with Waymo and the rest of the players is that they’re really moving into a lot of four-season cities, like Detroit. So next year, Waymo is going to be launching a service. And what we’ve seen so far is that a lot of these autonomous vehicle companies are launching in Arizona in the Middle East, where guess what?Grace Shao (1:08:29) Resources.Tu Le - Sino Auto Insights (1:08:55) Weather is super predictable, and it’s pretty one-note. And where the edge cases are, you can look at it like an inverse normal distribution curve, and I’m oversimplifying this, Grace, but the edge cases- so they happen few and far between- but they’re likely where the most severe accidents happen. Okay. So so it’s like there’s the least amount of data available. Exactly.Tu Le - Sino Auto Insights (1:09:21) Right. Severe storms, blizzards, snow, whiteouts. And so what we’ll likely see the final frontier being Robotaxis, because we’ll see commercial trucking from companies like Kodak. Remember that company like Too Simple? Those competitors. We’ll see those Aurora. We’ll see commercial trucking happen sooner. And likely highway to highway. I I divide commercial trucking into like three segments, and that’sTu Le - Sino Auto Insights (1:09:49) You know, intra-city, city to highway, and then highway to highway as three separate use cases. Yeah. And you know, robotaxis, I think it’ll be phased; it’ll be geofenced for a long period of time. There’ll be certain use cases that it makes a ton of sense for robotaxis, you know, but ultimately, in order for this service to become ubiquitous, there probably needs to beGrace Shao (1:09:54) It’s just much more predictable. Yeah.Tu Le - Sino Auto Insights (1:10:16) More services that have multiple people than one person in a car. And so that’s where it’s interesting because Waymo just launched the OHI, which is the Zeker contract manufactured vehicle, which has multiple seats. And I think that’s how Waymo looks towards profitability to get more than one person in an autonomous vehicle and almost looking at it like a bus, you know, a smaller bus to getGrace Shao (1:10:41) I was just gonna ask actually, like, are we gonna see a redesign of what robotaxis should look like? Because you, you, I tried out Waymo in San Fran, and it’s kind of creepy because you still have the driver’s seat, but like no one’s sitting there, so you’re constantly freaked out, like, what, you know, if you’re not used to it. So, like, will we see more like these little boxes or something that will signal to other drivers as well? More obviously, this is a robotaxi versus like a normal car.Tu Le - Sino Auto Insights (1:11:10) For sure, for sure. Right now there are policies in place that say you have to have a steering wheel, you have to have brakes. But as the autonomous vehicle landscape evolves, we’ll probably start to see, and I have a theory, Grace, that more and more cities will limit private passenger vehicles coming into the city center. Okay. If we look at Paris, they’re investing 300 million euros to make all the boulevards that lead into the Champs-Élysées bike-friendly, and they’re gonna limit private passenger vehicles. And so especially in Asia, I could see that being, you know, maybe you park, or you take the train into the fourth ring road, and then you take an autonomous vehicle into the city center. And then to get to your office, you take a scooter. Right. So there are these scenarios where I think more and more cities will try to take back some of the streets, some of the roads that bleed into the city center in order to lighten up traffic and take back some of the land. Because if we think about, and we’re getting getting off topic here, but I think these are important kind of secondary and tertiary effects of autonomous vehicles. Look at these parking structures and like these parking lots. We use them from like 7 a.m. to 5 p.m. And then they’re not used for the entire rest of the day. It’s kind of a waste, to be honest with you. And then and I’ll right. And so if we can take that back, you know, ideally make more housing affordable, make more office buildings, or whatever, right? Like round out, make more green space as opposed to having so much so many parking structures. Hong Kong could use more green space because all they can do is build up. And so there’s a lot of opportunity.Grace Shao (1:12:36) Hong Kong has, like, only I think like less than ten percent of the population even have a car because the public transit is so good. To your point, like there’s minibus, there’s a double-decker bus, there’s like the T MTR. Everything is walkable. you it’s and it’s c really, really, really well planned. And I think a lot of Asian megacities are like that. Actually, like think of Singapore, think about like Shenzhen, obviously obvious Shenzhen where you have likeTu Le - Sino Auto Insights (1:13:05) All right. Yes.Grace Shao (1:13:26) EV buses to EV cars to EV scooters, all for rent, all for access for the like average person on the street, right? So yeah, that does make sense.Tu Le - Sino Auto Insights (1:13:34) I’m gonna drill down on that.I’m gonna drill down on that because when I was living in Beijing, I was a mobility practitioner. I walked, I rode share bikes, I rode subways, I rode high-speed rail.Tu Le - Sino Auto Insights (1:13:53) Yeah.Yeah, well, I mean, rings around a road is a little weird, but now that I live in the United States, I’m just an advocate ’cause all I do is get in my car and drive everywhere I go. And on the weekends yeah, yeah, wellGrace Shao (1:14:07) But it’s also because they’re in Detroit. It it makesa difference, right? If you’re in suburbia versus like the middle of like ring three Beijing.Tu Le - Sino Auto Insights (1:14:14) Yeah, and I think that’s a big difference between like North America and Asia. And I would lump North America and Europe a little bit into that because a lot of these cities aren’t very big. So to invest in subways and things like that would probably be dis a disproportionate expense for the city’s budget. Where when you’re in China, there are dozens of cities with over a million people, you know, hundreds of cities over a million people.Grace Shao (1:14:41) Over like twenty million people, like a couple cities are over, yeah.Tu Le - Sino Auto Insights (1:14:43) That’s why I think, and what’s important about China or distinct about China as well, is that the automotive sector didn’t build out this transportation system because there’s a balance of high-speed rail, to your point. There’s a balance of subways, intracity transportation. But I’m off topic.Tu Le - Sino Auto Insights (1:15:07) The autonomous vehicle space isgonna be very interesting because our is the US government going to restrict silicon? you know, does because right now NVIDIA basically supplies every automaker with a a high-end transport or intelligent driving feature. Okay. But you and I know that the Chinese government is really pushing for companies like Horizon and and andTu Le - Sino Auto Insights (1:15:35) Black Sesame and XPeng, NIO, theirHuawei, they’re all silicon design companies now, too. And eventually NVIDIA is gonna get pushed out. And that’s also another bifurcation point as well. So if the Chinese can’t catch up to NVIDIA and Qualcomm and some of these other silicon design, Western, more established Western. Silicon design companies, does that mean their AI is not as good? Does that mean it’s not as robust? I think these are really open ended questions that you probably have conversations with your other guests on. So, and and I listen to you because that’s important to me of about understanding other perspectives on that stuff, because although I understand the chip sector, not to the level where I’m not an AI expert. And so I, like I said, I use it. as a tool as opposed to the end-all be-all. But but yeah, soGrace Shao (1:16:32) No, appreciate your insights. Really, really appreciate your time. I’ve definitely taken up more than, you know, I asked for. So, to end, I wanna ask a question. what is one differentiative view or you think a misunderstanding the the world might have of on the topic of China EVs, mobility?Tu Le - Sino Auto Insights (1:16:51) I think in twenty, twenty-five years, we’re gonna look back at this time as a renaissance in mobility because of everything that’s happening so quickly and being driven by the competitiveness of the China market. And we’re gonna see BYD is definitely gonna be a player. And you know, the other thing that I think is really, really important is that. I don’t believe traditional automotive folks can think outside of their normal way of seeing how the world works through transportation. And the top 10 mobility providers, to me, in 15 years, there might be a handful of traditional automakers, but I see an Uber maybe being a top 10 player. I see a Baidu or a Waymo being a top 10 player, and they don’t. build cars, but we’re the the importance of building vehicles is not going to be it is going to be reduced over time very quickly because of China. And so if you’re not providing a value added service in the mobility space, which I think which I think China is going to be able to do at at a much more affordable price point, especially in the emerging markets. I think that’s where the important thing is because in the Western markets, the BYDs and the Geely’s still have challenges and customer acquisition costs are much higher in in those emerging or those established markets. But in the emerging markets, the Chinese are going to try to roll out not only passenger vehicle buy-sell and their brand, but they’ll probably try to sell a lot of services once they have that sale. And I think that’s really going to be that opportunity for the Chinese to really make a name for themselves. Because you know this, Grace. One of the coolest things about the United States for me as an American is that anywhere I go, and you can not like the food, you can not like the coffee, but it’s consistent. If I go to a Starbucks in Munich or Vancouver or Toronto, and that’s soft power. That’s American soft power, right? The Chinese would love to have four Chinese brands. Creating soft power for them, creating aspirational desires to have their products. And and so that is gonna be the priority for a lot of these entrepreneurs that you and I speak with because, you know, they’re as ambitious as Elon, you know, maybe maybe they don’t get covered as much by Western media because their English might not be fluent or whatever, but they shouldn’t be underestimated just because they’re in China and they’re not in the rest of the world yet. And and I think that we’re gonna look back at this time and point to a few Chinese people at the level of being close to Elon. SoGrace Shao (1:19:47) Very interesting. Yeah, I think I think to your point, a lot of the entrepreneurs I speak to these days, they set their eyes on the global market in the first day and they want to set the industry standard. Like that is their goal. And it’s no longer about shipping out something cheaper, shipping out something just to make that quick buck anymore. So there’s definitely like a a sentiment shift and that confidence is different as well. to it Yeah. Well yeah. Thank you so much for your time today.Tu Le - Sino Auto Insights (1:20:09)It’s off the charts. Confidence is off the charts.Grace Shao (1:20:15) Really appreciate your time. I feel like I need to invite you back for another conversation because, you know, for what I prepared, we can go on for another two hours, I feel like. But it is late tonight, for me. So I’m gonna call it a day. Thank you so much.Tu Le - Sino Auto Insights (1:20:30) Thanks for having me, Grace.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Where does Europe fit in the so-called China-US AI race? 08.06.2026 1ώ 9λ

  Joining me today is Alex Lu, who offers a unique perspective. Alex works at the intersection of three very different AI worlds: China, Europe, and enterprise transformation. Having spent more than a decade in France and now advising European companies on AI adoption (often Chinese models), he offers a perspective that is often missing from the broader AI conversation, which is typically framed as a competition between the United States and China.In this conversation, we explore how European companies are actually approaching AI implementation. Rather than racing to deploy the latest models, many are focused on organizational design, employee adoption, process changes, and measurable returns on investment. Alex explains why European firms tend to be more cautious than their Chinese counterparts, how concerns around AI sovereignty shape technology decisions, and why companies increasingly find themselves balancing U.S. frontier models, Chinese cost-efficient models, and European alternatives such as Mistral AI.We also discuss the economics of AI adoption, including the emerging concept of “tokenmaxxing” or rather if that is even the wise path forward, whether AI is truly replacing jobs, how companies should think about ROI when AI introduces variable costs, and why the future may involve token budgets becoming as commonplace as mobile data plans. Finally, we explore Europe’s position in robotics, industrial AI, and regulation, and whether Europe’s strength may ultimately lie not in building the largest and best-performing models, but in defining how AI is deployed responsibly at scale.To find the previous episodes of Differentiated Understanding, see here.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.Season two will host a series of guests from early-stage investing, as well as builders, researchers, founders, early adopters, and product managers. For more information on the podcast series, see here.AI-generated transcript (for reference only)Grace Shao (00:01)Hi Sheng Yun. Thank you so much for joining us today. Really excited to have you.Alex Lu (00:05)Yeah, thanks very thanks for inviting me. I’m also very excited to have this conversation with you.Grace Shao (00:11)Yeah, awesome. So tell us about your journey. I think you’re in a pretty unique position. You know, like I said in the intro, you know, a lot of the conversation about AI right now is often positioned between China versus US, But you actually work predominantly with European companies in adopting AI and their digital transformation. So tell us about your your background and how you got into this.Alex Lu (00:31)Yeah. so thanks a lot. So actually, I went to France. I spent more than 10 years in France. I went to France in 2004 and I studied in a school called Ecole Polytechnique. and then when I graduated from the school, I started my work in in Europe, mainly for automotive industry and afterwards for the consulting industry. And still when I was in the consulting industry, I worked mainly for for the auto sector. SoI have a very traditional background of automotive. That’s why some of the work I’m doing currently in the in the AI, we can come back on that, is in the automotive manufacturing sector and mainly for European companies. Because I started my career in Europe, so I know I don’t I know them pretty better, pretty good. And the the the other thing point I want to mention is the school I started actually the Ecole Polytechnique wasLet’s say it it was a famous school in France or in Europe, but it it’s not so famous in in the world. actually this is in France they have a different educational system. but still with the with the rising of of AI in Europe, especially the French large language model called Mistral AI, the school becomes famous because the founder of the of of of Mistral AI comes from the the same school. So basically it’s also a a little bit likeTsinghua university in China is like the the Tsinghua in in France, having the best talents for for the AI. So nowadays, when I continue my work in the AI transformation for companies or AI implementation for the companies, I work a lot with European companies. Firstly, I know that my I as I said before, and secondly, is when we look into the global competition between China, US, and Europe.In the AI landscape, it’s pretty clear. It’s like China and and US or US China being the tier one or first ranked models. And Europe is kind of lagged behind. So most of the European European companies, they have this kind of attitude of being a little bit complex, I would say. on one hand, they are kind of seeking for, of course, for the best technology in the in the world to enhance their company’s competitiveness.There comes the question, how I can define my AI strategy for next year’s between Chinese and US tech stack in AI. And the second question they raised often is while we are European companies, we want to keep keep our AI sovereignty, which is a very important topic in AI. again, we can come back on that. So their question is: okay, between this US and China tech race.Is there any place for European companies regarding the foundation model companies or application companies or even corporate clients? What could be the playground for European companies? So these are major two questions are often received from European companies and you will you can see the thinking angle is they European companies want to at the same time keep it keep the AI sovereignty and at the same time keeping their competitiveness. That makes the question a little bit complex. Yeah.Grace Shao (03:49)Actually why don’t we just double click on the unpack that a little bit? What’s your view on it? Like what what do you advise your clients to do then if if they are kind of cut caught in a pickle or unsure how to build out the next stage of their infrastructure kind of being caught in between China and the US?Alex Lu (04:08)Yeah. So the the first thing I I always shared is in in in this tag race actually China and US we are not I want to twist twist a little bit the angle saying this is a competition between China and US. Actually, if we look into details, actually China and US are taking different directions in terms of the AI development, if I can say, because let’s say if we look into the US,AI ecosystem or the AI development. I think a lot of efforts are put on the foundation model or kind of foundational research regarding how AI can be become AGI can be bring beneficial benefits to the humanity, or how we can guide Rails AI so that okay, one day we will not go into the direction of science fiction movies. So this is a little bit the the push from the US AI companies. While in China, actuallythe ecosystem or from the national perspective, China’s AI is more about applications and more about how we can have the s beneficial from the whole society from the AI and how I can combine AI with my traditional technologies or traditional business to to to to to grab more values. So if we think in this angle, actually it will give us two different pictures. One is we cannot say that it’s kind of fromfront to front front competition, because these two nations are just take different angles. The second thing is if we look into details based on these assumptions, we will say one nation is pursuing having the most advanced AI technology and one nation is pursuing most kind of most beneficial AI for the society regarding cost effectiveness, et cetera, et cetera. So then it comes to the question that you raised for European companies isWe always brainstorm and conclude on the simple question is what kind of AI are we looking for for European companies? Are we looking for, let’s make it simple, I take some an analogy. Are we looking for kind of you need all the employees to be the PhD employees having the most intelligence in the world? Then that will be the US foundation models. Or if we want to say we have the most cost efficient and best performing employees, virtual employees in your company.Then we might consider Chinese models, foundation models. Then this is the trade-off. I think the companies should figure out. And the answer will not be so simple like that, saying, tomorrow I will switch to all US tech stack or Chinese tech stack. I would say the two ecosystem, as in the past in the digital area, will still continue for European companies, meaning that they need to juggle with Chinese tech stack in certain markets.maybe in Chinese market for sure, but for other markets, developing markets where the Chinese foundation model are taking influence and as well as with US models. So this is the thing. And I think the other angle answer to to to their question is I I usually take the statement from Jensen Jensen Huang saying the AI is kind of five layer cake.So what we are talking about is only one layer, which is the foundation model. And if we go deeper, then we will have infrastructure like data centers, like powers, chips, and electricities. And if we go upper, we will have the applications. So I would tell European companies or I told European companies often is I think the use cases in Europe makes a lot of sense because the cost is there and the employee was prettymuch expensive than Chinese employees. So if we deploy the same model, let’s say, and it of cost the ROI return on investment, you make the business case very easily in Europe than in China because the labor cost is kind of lower. And the the advantage of Europe, one I would say one of the advantages is about power and electricity. I study in France and in France y you would see they have the most advanced nuclear nuclear power technology in the world, at least in the past. AndI think the French government is also think about how we can build more power plants in the in the country to support Mistro’s AI development. And I listened to the founder of Mistral AI, Arthur Mensch. he explained to European Commissions how we can keep the AI development in the Europe is just he make a very simple analogy, meaning that intelligence equals to token. So we all know that, and he said token equals to electricity.So if we want to make our society more intelligent in Europe, then we need to build infrastructure and more efficiently and more sustainably. And and my last point is I looked into the report released by Stanford, the HAI index. And very interesting because US is far away beh in advance compared to other countries in terms of the number of data centers, it’s around2000 or more than 2000. I I didn’t remember the exact numbers. And the second and third, it’s not China in terms of the number of data centers. Based on the data, it’s United Kingdom and Germany. So Europe, Europe has the capability to build power plants, but I I I tend to believe these power plants are not currently used to train US models, in my opinion. So again,This is the European competitiveness if we want to talk about AI. So if we enlarge a little bit picture, we say that okay, it’s five or five layers cake, then again, China and maybe is better performing better in terms of electricity, maybe a little bit less performing regarding the chips. Same situation for Europe. So it’s not only about the most performing models, right?Grace Shao (10:13)That’s an a really interesting take. So help me understand like what do you actually advise on these companies for? So you gave me a big high level picture, right? Well give me some examples on the kind of work you’re working on. It’s because I think on this podcast, we often invite people who are builders, founders, investors, and they give us a lot of high level views, which is great, right? But but I want to hear from you, how are you actually helping companies go through this AI transition? AndWhat are the bottlenecks? Maybe further down we can talk about that. What are the challenges? What are the exciting areas? But just help us understand what are the day to day tasks that you’re working on.Alex Lu (10:49)Yeah, thanks. again, I I might share two different perspectives from my experience when it’s again with the European companies. It’s very interesting example because actually I build a product doing this kind of market intelligence, market research for European companies and the value proposition at the time it at that time is we can save time for your employees and they a and and we make your organization more efficient.And basically we find when we when we s when we sell this kind of value proposition to different companies, I see very interesting different answers. One is on the European side, he would say, this is very interesting, but before implement implementing, we need to think about a kind of tomorrow’s process process, meaning that if we put your AI product into our organization, how our employees will work with the product together, what’s the process look like?And how many new skills would my employees need to perform or to better use your products? So I think the European company’s mindset is they will need some time to conceptualize the AI products or AI use cases. And then they will need to conceptualize and project, especially once the product is in place, what my company will look like. So they will spendA little bit more time than Chinese companies to figure out the the regarding the talents, regarding the organization, regarding the process. And in my opinion, it might be a right right approach, in my opinion, meaning that they put human before the techno technology. And this is what I observed when I implement AI for companies saying that, okay, I bring you the best technology, but often we might have improved efficiency by 10 times or five times.In one single process, but actually there will be some bottlenecks in other organizations, in the in the rest of the organization, then you cannot you cannot increase the efficiency of the whole workflow, let’s say. So that’s why a lot of people in US they talk about AI native organization to kind of remove the bottlenecks in the in the organization. And I worked another example, I worked for a European company, and it’s very interesting. He said, I receivehigh level management. He said, I received so many reports from my employees, I I don’t have enough time to review and to approve them. That’s the case because we increase efficiency of the working level of the people, then the bottleneck becomes suddenly the a the leadership. And then we need maybe an empowered leadership by AI in the future to make the whole organization more efficient.Or we need to think about a new organization where we include AI agents and human beings together because the two natures are producing things on a different scale. So this is mo most of the time, this is the European companies. And for the Chinese companies, the mindset is totally different. if we implement the AI solution, the same product to a to a Chinese client.The the the answer would be, that’s very interesting. You save 20 or 30% of my employee’s time, but you know I cannot let’s say lay off the employee and to make some savings. So just tell me for the time we saved where he can work to produce more. So it’s always in the mindset, okay, we have some some time saving, but you I cannot pay 80% of the salary to the same guy. SoIn in order to so I I need to pay him hundred percent salary, so I y your business case doesn’t work for me. So where where we can grab more values. Yeah, so you see a diGrace Shao (14:37)That’s really interesting. That’s a really interesting approach.Yeah, because it’s like the company is reflecting actually a broader, I think, social, even cultural and perspective on how they’re perceiving AI. And in the China and US, often the conversation is so fixated on improving efficiency and people who are utilizing AI are actually more burnt out because they are like 10Xing themselves or whatever these days.Alex Lu (14:49)Exactly.Grace Shao (15:03)But you know, in Europe, that conversation is so different. And you can say maybe much more humane. However, like you said, the bottleneck right now is then how do these companies become the next generation? Like still relevant in the future, once this becomes normalized. So it’s interesting. I wanna go back to that a little bit later as well. I I wanna touch on something before we get further into I guess the comparison of you know, the adoption and everything isAlex Lu (15:17)Yes.Grace Shao (15:30)You and I met each other essentially online because I found out about your work that you helped a lot of European companies adopt Chinese model. I found that was very, very fascinating, right? you advised them on how to basically integrate, say, the Minimax and C AI of the world. Now, a lot of these model companies, when I speak to them, they say their priority right now is to basically sell globally. And of course, the Western markets are some of the most lucrative markets. the US.headwinds are mostly in geopolitics, compliance, Europe. How do you view that as a market for them or opportunity for them? Like, is it equally challenging for these companies to sell to enterprises in Europe or do you think there are more opportunities for them right now and they’re they’re kind of taking off a bit more?Alex Lu (16:16)Yeah. I I I I if I think the conclusion if I I can state at the very beginning is kind of in Europe definitely there are more opportunities than for Chinese large language model companies than in in Europe, than in US, sorry. I have two proofs for that. Firstly is I discussed with a a CTO who is also a schoolmate from my school Polytechnique.And actually he was very interested also by my newsletters on LinkedIn and then when they he asked me the question so apart apart from Deep Seek, what are dip other models and Chinese models that we we that we can use to improve our efficiency? Because when we meet all of he said he told me when we meet most of the IT implementation companies, they came with the solution like Anthropic or ChatGPT or OpenAI or or Google Gemini.So we don’t see so many options. He mentioned the word options of Chinese models. And we know that Chinese models are more cost efficient. And then we can talk about token mapping. I think it’s kind of related topic. So this is one thing. I think in Europe, actually for the companies from the business perspective, they are also looking for different variety of different models so that they can bring what I said before, a best cost performance ratio models in the in the organization.So this is one thing. And then I I told him that most of the Chinese large link model companies, firstly they started their business in China and then they tried to inf have the global influence. Like the most advanced one is zero dot zero one dot AI, but the other ones they are trying to catch up, like that AI you mentioned also Minimax. so I I think the thing is what I see today is the ecosystem of Chinese models.Are not currently penetrating into the European markets. But definitely there’s a a room for Chinese players. The second thing is I always take the comparison with the other industries like EV industries, like car industries. because you you will see Europe put a lot of tariffs on Chinese vehicles. because okay, you you see a lot of Chinese vehicles because ofEurope wants to protect their own industries, et cetera, et cetera. But at the end of the day, they are not putting hundred percent tariffs. They are putting somehow reasonable tariffs on the Chinese vehicles. So the bottom line I want to mention is I lived in Europe before and I know the mindset of Europe European people. The mainstream, of course, we have different views. I think the mainstream for European people, most open ones.Are saying okay, we need fair competition. The EV cars is just because okay, the European commissions are claiming that okay, you produce in China, but we in Europe we produce in a more sustainable way, so our cost is higher, blah blah blah. So if we take this comparison, I think definitely there will be some places for Chinese companies in condition that we play fairly in the European market.And then we might come back to the third point I mentioned before, of course, there’s a a point of AI sovereignty. the biggest, the biggest player of European AI ecosystem is still Mistrol, so it’s the biggest player in the foundation model. And of course, Mistrol should be one of the choices options when we suggest to European clients as the large language models.So I would see that if tomorrow the Chinese model enter these European markets, they will face a fierce competition with Mistro because Mistro basically they have a government back, let’s say, from France, and they have a very good positioning in the ecosystem. you would see in in two or three weeks you’ll there will be a VivaTech in France and Mistrol for sure they will be on the stage and for sure they will beFrench or German presidents, French president and German chan chancellors. And with their unique positioning, I think most of the European companies they were firstly considered Mistrol, but still if Chinese companies can bring something on the table, business wise, the European companies will not only limit to only one model, there will be some balance between different models. And today, the balance I see is Mistral versus other US models.Grace Shao (21:11)No, I just think it’s really interesting ‘cause I think it also totally makes sense when I to talk to people who are in the Korean market or, you know, covering the Middle Eastern markets. sovereign AI is just such a top of mind like conversation for companies, whether it’s for compliance reasons, or regulatory reasons, whatnot. So it makes a lot of sense that Mistral’s position very well in Europe. However, are there any other players that maybe we’re overlooking outside because we’re not that familiar with the European market? Any otherfoundational model labs that we should know of coming out of Europe.Alex Lu (21:45)Yeah, there will be apart from Mistro there’s another large language model whose name is H, but it’s less famous. And then you’ll have it’s not if we can say it is kind of word model by Yen Laquen, the ex researcher in Metafair, and he just came back to France and lab raised raised a large amount of money for the for for his model. It’s called AMI, yeah, AMI.Grace Shao (22:01)Mm.Interesting. Okay, so let’s talk about the token maxing thing you touched on just now. So offline we talked about this a little bit recently. There’s been getting some buzz. It’s quite funny, you know, whether I’m it’s like big tech in the US or big tech in China. When I talk to them, people are saying, Okay, our managers are pushing us to token max. If we don’t basically use AI in our job and figure out ways to essentially replace ourselves, we get replaced, which is the irony in all of this. It’s it’s all kind of sci fi. butGrace Shao (22:41)Then the joke’s kind of been played now on the companies because you know there was just headlines coming out saying, this one guy basically spent like more than half a million dollars on tokens in a month, and that’s obviously more than his salary. And then companies are realizing, wait, this token maxing strategy is not cost efficient at all. So from an operational standpoint, I know you are someone who work a lot with companies to implement AI and findAlex Lu (22:54)Yeah.Grace Shao (23:10)the most cost efficient way for their for their operations, right? And not just costs, like you mentioned, it’s like a balance of costs, you know, and and operational sustainability as well as obviously company morale and everything. So how do we view this trend? Where is this going? Is this sustainable? Like just just give us some high level views on this.Alex Lu (23:32)Yeah, there there’s a a lot to talk about this, because the token is becoming really a trendy topic for individuals and as well for companies. so to answer your first firstly to answer your questions, I don’t think that’s sustainable. My view is the token mapping is kind of marketing for infrastructure companies. and of course, as you say, there’s a lot of people burn a lot of tokens and more than their salaries.Then the question would be if I pay your salary or if if I pay your tokens. We’ll come back to this point afterwards. I discussed with some some Chinese companies. Very cost cons cautious. I think the the the thing is today when we actually for for the tech companies in China, there’s also some ranking of token consumed. but it’s kind of indicator of how people are use AI. ButIt’s not is it the right indicator? I don’t think so. basically I think in the in the in current status we didn’t we didn’t find a very good metric to measure the performance of a human being empowered by AI. that’s the thing. So we take a kind of proxy indicator, which is the token for and of course there’s a lot of waste of token in in in in in the usage and I’m I’m not sure that every single employeeswould be the master of AI if we don’t provide the sufficient upscaling in terms of the AI. Because from individual perspective, sometimes we use by coding, but if we don’t master the basics of coding, then we might waste some time and as well as some money and tokens in the by coding. So this is my view. So the token maxing is kind of marketing stuff and and the the day when we find outAgain, for the AI organization or for the organization, how we can measure the performance of individuals with AI, then we might have a clear picture and no longer token max. And the other interesting thing you you mentioned already, but I read also is Microsoft they are kind of switched to their copilot because ever s if everyone used used the entropy cloud model then become too expensive for the whole organization. It’s just not just not cost efficient.And brings me to my point is when discussed with some Chinese companies. So, you know, Chinese companies are very cost conscious. And they are thinking is I think that it was a joking, but this is right angle of thinking is show we in the salary of our employees to allocate a part of the tokens monthly for our employees. meaning that okay, if theIf in the in the in the past situations hundred percent of the salary tomorrow might be eighty-five percent of yesterday’s salary plus fifty percent by tokens. And the tokens you can you can use and if you don’t use tokens efficiently then it’s the savings for the company. So this is it’sGrace Shao (26:43)That’s really crazy. But I kind of see what you mean.Like so essentially it helps you with your job. So that’s why it’s on you. But then what if you just don’t w but what if you don’t want to use AI? What if you just like I can do my job perfectly fine the way I did it before and I don’t want to token max and I want to keep my hundred percent?Alex Lu (26:50)Exactly. That’s the question that the Chinese company needs to answer, but you reflect on your point mentioned that the token consumption is sometimes much more expensive than the salary. So it causes Chinese company companies to think that okay, I spent salary, I spend tokens for the intelligence, I spent two times to hire employee. So why not combine them together and doing kind of tomorrow’s package is your basic salary plus tokens?Grace Shao (27:32)So actually on on that,how should companies think about it then? Because, you know, it’s really easy to say, okay, this is an AI native company. There’s 20 people in this company. Everyone’s token maxing because it does bring the 20 people’s efficiency to say like 400 people, whatever it is, right? However, what about the traditional companies, especially the ones that you work with? Like a lot of them are OEMs, manufacturers, you know.It it doesn’t make that much sense for them to really jumping on this AI bandwagon as well then. Or how do you advise them then? Or how do you think how should they think about it?Alex Lu (28:06)Yeah. I I think for the for the traditional companies or European companies, it doesn’t make sense for everyone to give the token maxim because as I said, I’m pretty aligned with the European approach saying that okay, in order to release or unlash the value of AI, we need at least to upskill a little bit our employees. We cannot expect employees like with thirty ex years experience in the industry and tomorrow he switched to a kind of AI expert in thein the in the in his company. So I I just want to combine our question with my previous comment saying that today if you look into the Chinese market today there are some big big traditional telecommunication companies like China Mobile they are proposing the token plan for individuals it’s like your smartphone monthly monthly plan yeahGrace Shao (29:02)Wow. Like data plan.Alex Lu (29:05)It’s a kind of data plan, exactly. So the token is becoming kind of infrastructure like electricity, like water, or like your smartphone, monthly subscription. So this might be the way the companies might pursue, saying that, okay, yesterday I might give you a kind of monthly plan for your telephone. So I can reach out to you and you can read the emails and you can use the telephone to walk with emails, work teams or with Zoom, etc. etc. And tomorrow it might be a com kind of monthly subscription.For different employees, then you have a monthly token plan you can use for your personal, not for professional work in AI. I guess that might be the way that the China might be moving for individuals and for companies. And again, for European companies, they are not there yet, but when I discuss this vision and this kind of trend, and they are pretty interested, they might be moving in the same direction.And for the companies, at least not at the national level, but at the company level, to provide kind of a monthly subscription to a limited number of people who master AI, and the first wave of people adopting AI is their coding team, their IT team, their digital team. So they will be the first employees to use this kind of concept of monthly subs subscription to tokens.And of course, for manufacturing companies, there’s a lot of people working in the factories, in the plants, or or in the on the production lines, and they are not be impacted, they will not be impacted by this kind of AI wave. But still, I think the things are are moving slowly and it’s it’s changing so quickly. but this is currently my discussion with European companies.Grace Shao (30:48)That’s actually very interesting. I it makes a lot of sense actually to build it in in as like a infrastructure like 5G data. And then it’s really, it’s really like there’s a cap on how much the company will pay for, but then how you utilize it should be, and you’re more mindful of how you’re utilizing this, right? And not wasting the tokens and and buy the that thus you know, wasting your energy, compute everything. SoGrace Shao (31:13)I want to bring it back to the Chinese pricing models really quickly. I know you work with a lot of European companies, they are the buyers essentially. You also help them connecting with the Chinese vendors, essentially, which are like the Chinese LLM labs, Minimax, Drupal, Moonshot, etc. Now, how should we understand the pricing model of these companies right now? Because it’s obvious that they are pricing themselves much cheaper to US peers.Some might you know, obviously argue that their performance might not be as on par like on par or as at the frontier. however, even when they do play catch up, you know, the reflection of it is it just s seems like a complete different cost structure. Help us ex understand that, like they’re thinking, why they’re pricing it much lower and how that plays out in the long run.Alex Lu (31:46)Mm-hmm.Yeah. Actually, there are two perspectives on that. maybe I will firstly talk from the client’s perspective and then I I might conclude with the recent price decrease by Deep Seek. maybe you you have already read about it. so f from the European companies as I said before, the thinking is w that the that that’s that’s the statement for the companies I met. we do not need entropic models for all the time.That’s for sure. Because this is very expensive even for a company. So for sure they will need a kind of different options from different models, like the best U US models and the cost-performing, the best cost-performing models from Chinese models and the AI sovereignty models like Mistro. So basically there will be three combinations and then there will be engineering of technical issue that meaning that how to manage these models toPerform the right tasks. So, meaning use cloud to perform the most complex tasks and use Chinese models to perform kind of less complex tasks. And I think European companies they understand this. And they, of course, they are looking for Chinese models for the cost effectiveness. And I would say this is also one of the bottleneck of US models because they are veryIn a relative way, very expensive. Therefore, it it’s the bottleneck of the massive adoptions. Only the European big, big companies can afford like continuous use of US models. well there are a lot of SMEs in Europe. So this is this is the thing. And for the Chinese model suppliers, I think the the way I I see the the the price issue is if you ask meCan Chinese companies increase their token prices? I would say surely, because if you look into the financial report of ZAI or Minimax, actually they are not they invest a lot in the research to develop these models. And the expectation from the industry AI industry is if you want to train or pre-train a next model, you will cut it will be more costly than the previous pre-trainings. so for sure.Chinese companies can increase their token prices. And w that that’s what they are doing actually after the open cloak, if you read into the news. And the thing is, compared to the US model, still the Chinese model are very cheap. I think there’s one very strategic thinking thinking angle is if you think about the Chinese models, most of them they are open source models. And the the the the thinking angle is, I think, for the Chinese model players isWe want we open source these models because we want people to use these models. Because they can deploy it on their own infrastructure, they will have more freedom, or they can use our open source model to train their own models. and maybe they they will use our our tokens by or they will they will understand or know our models better by open sourcing. So if if we combine this thinking angle, I would say.The Chinese model strategy might be to increase the influence in the world, maybe in the developing markets, where people are more cost conscious, and to help people to use this AI to adopt AI in a cheaper way. And then in the long term, in the future, that’s very Chinese, maybe again to increase the prices once we take the market positioning.It’s like the price competition for the last decade regarding this digital sharing economy or digital era. Nothing has changed. So a very aggressive c pricing strategy to at least to to have the market share and then once we have the market share then we can establish our our our our position in the market and ca kinda do a lot of monetization stuff.That’s the one thing regarding the increased influence globally and taking the lead in the AI industry for the developing countries, in my opinion. Of course, go going to Europe is is part of the their strategy. So this is from the Chinese model’s perspective perspective, and it’s a very special case, of course. It did this is Deep Seek. DeepSeek released just the before and right after the release, during one month, I think for the developers we enjoy the75% of discount regarding the token price. So it’s very deep discount. And recently, I think one or two weeks ago, DeepSeague announced that they will keep this 75% discount for for for forever. So it’s kind of they they just discount their token prices by such huge amount of discount. I’m pretty surprised. andAgain it di it it launched a price war in the market and you see recently Xiang Mi decrease also their token prices and I don’t know if other players will will follow in Chinese market at least. But if we think about Deep Seek cases, it’s a very special case because Deep Seat this year it doesn’t create a lot of buzz in the AI community in the US. I think so. I I’m not living in US but I read some newses. news, sorry. I think thenowadays DeepSeek, I’m not saying that we have the best performing model. and and and in terms of of the tok coding performance, DeepSeak is is not at the top top level compared to other models. But the interesting thing is DeepSeag this time is trained on the Huawei ASEAN chips. so again, I think the price decrease of DeepSeag combining with their recent news of raising money and hiring some harness engineeringAcross the world, I would suspect that DeepSeek by decreasing their prices, they just want to break through the ecosystem established by NVIDIA. This is my thinking, and and that’s why after the President Trump visit to Beijing, there are 10 Chinese companies are not authorized to buy Nvidia chips, but up to now you see few others.Grace Shao (38:09)That’s interesting.Alex Lu (38:23)I think there’s a thinking from the national wise from from the nation thing that okay with Dipsy can we break through the Nvidia chips plus CUDA? And if because that’s so cheap, so most of people they might use Deepsi in the future and they might be used Huawei as ASN chips because Deepsi got trained on these chips and it’s best support DeepSeak’s performance. So this is another angle. Yeah, so you would seeGrace Shao (38:23)Mm-hmm.So the open source strategy. Sorry, go on. It’s basically a strategyto get people in to get the developer into its ecosystem, its own community first, which is what Jensen’s been saying the whole time. Yeah. no, I I agree with you on that. I actually I I wanna and steer away from the chips today because I I am quite fascinated. So you work with companies, adopt AI, but how does that actually what are companies really using AI for? Like we hear about stories.Alex Lu (38:54)Exactly.Grace Shao (39:16)you know, companies are token maxing, whatnot. And obvious the obvious one, like you mentioned, is in coding capacity in IT, but no again, not every company is in tech, you know, not every company needs coping co coding capacity. sorry, let me just say that. Not every company needs coding capacity. So like what are we seeing actually on the ground, especially for maybe more brick and mortar stores or old school traditional industries? Why would people all want to adopt AI right now?Alex Lu (39:47)the the the adoption rate actually for European companies is pretty low, to be honest. most of companies, if we say at a large scale, they don’t adopt sufficiently AI and they just are afraid of missing out something. So this is a FOMO. they are just feared of missing out some opportunities, and if they don’t use AI today, they might be less competitive in the future. So the the f the most common use cases I see incompanies for coding and for it and sometimes it’s easier to measure the effective effective sorry effectiveness of ai that’s in the most most of the time in the sales marketing department so meaning that if you use ai you can produce produce more contents and with more contents you have more impressions with more impressions you might have more conversion rate you might have more conversions and you might have more sales revenues soThis chain is actually well formed. So by using AI, you can track the individual metrics on the chain, and then you can kind of monitor the results by using AI. And most of the time, I get a very simple question of European companies, and very difficult question actually to answer is: what’s the ROI of implementing AI? What’s my return? then it’s a very difficult question because in thedigital, 10 years ago in the digital era, I can tell the ROI, I can estimate why, because the incremental cost of using digital products is kind of almost zero. You just need your digital products and then it makes more efficient, it makes more automate. Well, in AI, that’s very difficult because if you think about it, if you use more AI, you will consume, as you say, more tokens. So, meaning thatAn employees, you need to pay the salary. If he is a heavy AI user to produce more content, then you will need to pay his tokens bill. And then the ROI might not be so immediate. Or there might not be ROI actually for the individual use cases. Then we come back to the question: is okay, by using this AI, how we can make the whole organization more efficient and how we can generate more revenues for the whole organization.While for the individual users, maybe there’s no business case. So I think again, the the the the the difference compared to 10 years ago is the people who use AI and who use heavily AI, then he will have a bill to to pay. That’s a variable cost. That’s very important. And secondly, is the variable cost will be reallyThe beneficial of the variable cost will really depend on the skills of each individual. You may pay $100 for employee A or employee B. If B master better AI, then you will have 10 times more results, financial results, compared to the first case. So again, I think you asked the right question. the ROI question is definitely a very good question. and most European companies they seek about ROI before investing. So they are very cautious.While again, if we compare to the Chinese companies, we are more pragmatic. So let’s implement a POC. it costs a little bit, but let’s implement it. If it doesn’t work, never mind. We waste some money, but we we we continue, we iterate or we continue with another use cases.Grace Shao (43:17)So you think the Europeans are taking a more cautious approach, but actually more cautious on what the potential ROI is. Then I bring it to the question that is a bit more philosophical and like a societal, not so businessy, is then isn’t the headline or the mainstream discussion on AI is replacing our jobs completely overblown then? If companies are not even investing in the like, you know, buying tokens, I don’t think they’re replacing people and comp just replacing roles with. Like AI, are are they? How do I understand this?Alex Lu (43:50)For the tech companies, I think your statement or the statement is true for the tech companies because they’re traditionally there are a lot of coders, there are a lot of programmers, and and actually I see a lot of developers, individual developers in the market because they work for tech companies and now with the with AI. That that would be very challenging. And again, currently for European companies, if I would sayThey’re still at very, very early stage compared to to China. the cost is one thing, and we can take at the other angle, causes equals to conservative. So they are a little bit conservative and they care a little bit more about their employees. So actually I I I will not see in European market AI replace a lot of human workers. It’s not happening today. Will will that happen tomorrow? I think so.Grace Shao (44:46)Mm-hmm.Alex Lu (44:49)but again we need to find another society structure or we need to find other job opportunities for the human beings when AI comes to the companies and replaces some of them. It we’re not like very aggressive like at the tech companies like Meta or other tech companies. it will happen slowly, but of course AI has impact on the on the employment on employment, even for European companies. andGrace Shao (45:13)Mm. The economy itself will evolve and and jobs will look different.Alex Lu (45:21)Exactly. it that that’s exactly what I I was in Europe ten years ago. It’s exactly the discussion around industry four point zero if people remember. We say that okay tomorrow we’ll have some automated machines in the plant. So it’s kept it’s not it’s happening currently in China. We call it a dark light factory. So it’s very automated. you can run the factory without turning the light on.so basically at that time in Europe we had a very big debate on where the employees employ employers should go once industry four point zero is in place. And the answer was there were sorry, the answer was there will be some upscaling and new job opp opportunities created with industry four point zero, and we need more skilled people to master these machines. And that’s that’s the same thing for the AI.Tomorrow we will need people who can orchestra, who can manage the agents, AI agents, instead of doing the same job as a simple agent.Grace Shao (46:23)Yeah, I see. So so on that, I wanna ask, you know, given Europe’s strength in industrial, like industrial strength manufacturing, where do we see opportunities for companies to really couple that with the development evolution of AI right now?Alex Lu (46:41)You mean the the use cases, right, for the companies?Grace Shao (46:44)Use cases, new opportunities, new potential businesses. where could we see p like, you know, new businesses come out or, you know, new business revenues for current industrial companies?Alex Lu (46:55)Yeah. for European companies currently the use cases we’re discussing is more around kind of efficiency use cases. So for example, they want AI to help them to do some root cost analysis because if you run a a plant and if the machine is kind of done, the production line is kind of stopped, and then you you you lose basically a lot of money because you missed up.opportunity of producing X unit units of of your products of your cars. So basically people care a lot a lot about how I can analyze the root causes of of a machine being done. And this traditionally was a very heavy task. We need we need a lot of experts to be involved and because there’s a whole system of different machines in the same plant. And the machine is kind of the product production line is kind ofmade in a industrial sequential. So every parameter on different machines might have an impact on the chain. So we need to involve a lot of experts and by using AI actually we can we can understand better. We can do some causality analysis and do some root cause analysis and find the root causes more easily and in the future to do some predictive predictive maintenance and to improve the efficiency of the companies. So this is currently happening.People are asking for that. And some companies are also asking for these kind of knowledge management platforms. Like we we need knowledge management for new enrollment of employees, for HR policies, for reimbursement policies, for new employees onboarding, etc. etc. So a lot of around that. And if we look into the vision and into the future, I think European companies are start to think about it.I’m talking a lot a lot about European companies, but that’s the same thing for for the companies in China, it’s just kind of more advanced. So sorry, I I’ll come back. So if we take into the vision of European companies, actually they are also thinking about the future, which is how I can use AI to increase my revenues and to make the pie a little bit bigger. And then it comes to the discussion of agentic economy.Meaning that can I use my agent to kind of sourcing, to kind of sourcing for my company? Can I use my agent to do some business development, to write emails, to do some code calls, to reach out to potential clients? So these are the things that people will come to think in the next wave, saying that okay, if we have a very good engineering of our agents, guidelines of our agents, what an agent can say, what he cannot say.what he should say in which context. So once this is done, again it’s very European, they need to use everything kind of under control. Then I think we are ready to to go for the athentic economy so meaning that agent can do business in in the place of the companies.Grace Shao (50:04)I see. And if I were to say I’m the founder of AI native company, how would you advise me other w because it would be very different from what you’ve been saying about advising more traditional industries?Alex Lu (50:10)Yeah, it it i if you are a AI native founder, I think I’m I’m doing the currently the same position. there are a lot of things to consider. For example, in terms of the technology, the foundation model is evolving very pretty quickly. So how I make sure that my AI agent idea or concept or business model will not be revolutionized or disrupted by thisFoundation models. This is something we need to think about. The second thing is I always tell myself and also people in the say same AI community is we we don’t start to build our products from scratch without discussing with the clients. So why not in in a more safer way, why not discuss with the clients, build products for certain clients, and then kind ofConceptualize the products and build more standardized products that we can sell, we can say, we can sell to market and we can scale in the future. It means the build of the product comes always from a specific demand of the clients. And once if there’s a demand, then we can do something, we can build things. Why this? Because, in my opinion, all the AI native funders, I think we are pretty aligned is produce.something or build a product in the future will be much easier in the past. And if we compete with AI in terms of the intelligence, there’s no way a human being can catch up with AI. And we should place our time where the AI cannot compete and where we still need a human being. I I I make very simple analogy to some friends of mine saying emotional intelligence, meaning that how we can establish relationships with the people, how we can build a trust.So still I think if I’m a founder or if AI native founder, he should go out to meet clients, discuss with clients, build a trust and have some demands from the clients because building the process will be pretty easy and the cost of failing is pretty low. So build fast, fail fast, scale fast and it works even more in the in the future.Grace Shao (52:36)And then my question on that is how do we actually understand how to build guardrails and safety around this? Because you talked about how Chinese companies you work with are often a bit more like gung ho, let’s go, we’ll t we’ll fix it if after it’s broken, kind of mentality. Whereas the European companies maybe are seen as a bit of a slow adopter in many ways, you can say more cautious, more humane, and protecting their concurrent employees. But, right, likeEnd of day, if this is the future evolution of our economy, how do we go forward with this? And then how do we actually build more intentionally?Alex Lu (53:13)Yeah. technic technically, actually there are a lot of skills, there are a lot of technical stuff in the area to build the guardrails for the agents, like Anthropic, I they are doing doing a very great job, and also some Chinese foundation model companies and also agentic companies. So all of all of that they call that the harness engineering. So they put every concept into the harness saying that okay, we need to build a harness and to make the guardrails.So this is the technical perspective. But still, this technical perspective is very from the developers or programmers. And if we bring the case into a real company case, then it really depends on each use cases on each company. I would say for any new human employees which is who is a new hire in the company, at least when I join European companies, there’s always a code of conduct.You see, it’s it’s simply a a document that we need to learn. We need to we need to we need to be compliant in the future in in our work or professional work within the company. So I would say for the AI agents that the same thing. they are very important in the future, a kind of infrastructure to evaluate the performance of the AI agents, meaning that if the AI agents is delivering the performance as we wished before, so there’s a kind of benchmark evaluation.And also the evaluation should include also is the AI agent performing correctly as we wished in terms of the code of conduct. And the code of conduct should in my opinion, be written by human human human beings. It’s like an extra bic team, they have they have written a a hundred-page of constitutional constitution for for for for cloud. And then each company should write their code of conduct for.every agent in every department. And a lot of Chinese founders then they are entrepreneurs, they are also joking at okay, we develop an AI agent today for companies but the next question will come shortly is when should we retire our AI agent it if it doesn’t perform correctly or why when we should replace them. So you see the evaluation or benchmark of of the AI agents wouldshortly become a a a pro a p a problem in the market when we adopt massively the agents.Grace Shao (55:45)So then each organization will have to institutionalize this, essentially you’re saying, and have their own standards of code of conduct, whatnot. That makes a lot of sense. Yeah. And right just like how companies right now regulate data usage, even company devices, whatnot, right? Like this will all just be part of the compliance that employees will have to learn. I want to ask you one last question, which is what’s one differentiative view you hold?Alex Lu (55:54)I think so. In terms of the AI?Grace Shao (56:16)In terms of everything, it’s a question I like to just kinda throw throw it at people when they come to the podcast. It’s a it’s a wild card.Alex Lu (56:24)Okay. I think one of the points I always mention, it comes back to my background, is today the AI race is between US and China. So we say that European is kind of lagged behind. but do not forget that actually technology is one thing and the usage of technology is another thing. And again, if we come back to ourmy my statement saying that implementing AI is not about technology. It’s not it’s about process culture and organization and human being. So I think the placard of the Europe is they’re pretty good at regulations. And if you think about they issued GDPR before the Chinese PIPO, which is protection of personal data. And they have this kind of European AI Act. And then if I think about how anthropicThey penetrated these enterprise solutions versus ChatGPT and generate today more revenues than open AI in terms of AR, because of the simple concept of responsible AI. Then I would say tomorrow, if the AI comes to the enterprise level, enterprise implementation, and if everyone should be responsible in the company with their own agent or with their own developed AI, maybe Europe has a part to play in that.in the in the AI in the in the world of AI, because their initial statement is kind of we want AI to be regulated, we want AI to be responsible. So this is my point of view.Grace Shao (58:06)Thank you so much. You know, today you’ve been really generous just explaining to me and and the audience just how AI is really being implemented into these big companies and the more European perspective. is there anything else you think we’re missing or any misconceptions we might have about the relationship between European companies and Chinese companies or how Europe is perceiving AI? Is there anything you think we’re missing or do you think we covered it all mostly?Alex Lu (58:36)Yeah, I I I think we covered most of them, but I just want to mention one thing is even though we say that okay, there’s two different nations in the world, US and China, competing AI, or in we we we take different directions of AI. And still I received a lot of recently questions from European companies, and they are really, really interested by Chinese tech companies. So you would seethey are pretty open and they come frequently nowadays to China and they have the mindset of learning what Chinese companies are doing, what Chinese foundation models are doing, and especially seeking their use cases. so one thing I would say is when I receive them, we show some very advanced Chinese use cases. They would say, you are in a different environment because we have different laws, we have different regulations compared to you Europe.but they are quite interested about what’s happening in Hong Kong because the regulations in Hong Kong is pretty closer to European markets. So still, I I see we might have a lot of potential collaborations between China and Europe in terms of the AI, in terms of the physical OI. We didn’t mention the robotics, and definitely it’s an area where European can have more playground, not onlyAbout the humanoid robots, they want also to have their places in the hardware value chain for the robot robots. Like a lot ofGrace Shao (1:00:10)I’m sorry, it’s I know we’ve hit our time, but what what is your view on that? Because you know, European companies traditionally been the leaders in robotics, right? Industrial robotics, like machinery. where do they stand now in the world? You know, are are the Germans and the Japanese still leading the space or or how how are they gonna be kind of presenting themselves or positioning themselves on the supply chain right now?Alex Lu (1:00:15)No worries. Yeah. So for the very traditional industry robot robots that let’s say it’s like KUKA, you have a lot of robotic arms. So they are still kind of leading the world, so you have a lot of robotic solutions implemented in the in different car makers’ plans. but for the humanoid robots, actually Europe Europe is lagged behind again because it’s not all only about the valueAbout the not only about the supply chain of the robots itself, it’s also about again the software and the large language models behind the robots. So the mindset of European companies today is: okay, we understand China again has the most advanced humanoid robotic companies in the world. US has maybe advanced in software in large language models or word models. China is pretty good at the supply chain.So again, the same question they ask themselves. But the the recent demands I receive from European companies are are two. The first one is as a traditional European companies, we know that they know that the value chain of making a car is quite similar. Let’s say it’s not hundred percent the same thing, but there’s sixty or fifty percent are common of making a car and making humanoid robots.So their thinking is okay, can we participate in the wave of these kind of robots with the development of China? So like motors, like electric motors, like actuators. Yeah, German, German guys are pretty good at at this apply. So that’s the first thing. The second thing is demand is a lot of European companies saying that okay, we have the real use cases in Europe because we are lack of workforces in our plants.It could be an aging population, it could be some strike of labor unions. So they in order to keep the plant working, as we said before, about the predictive maintenance, they are very welcome, the Chinese robotics in the European markets. Again, the robots need to be compliant with European regulations, conditions, and they are very welcome. So the most common demand I receive is hey, hey, I I want to do a kind of analysis abouthow I can be part of the supply chain in China and how I can leverage Chinese supply chain to be more competitive. The second one is okay, I have a use cases, then we need to think about how I can implement the humanoid robots in the European markets. And then we we can discuss about the business model of the robotic companies like Unitree of AJ Boss, because it’s not only about putting their robots in the factory, it’s about calibrating the robots, it’s about capturing the data, it’s about think about a closed loop of robust training. It’s aboutthe again, the guardrails how make sure robots will not harm a human being if they cross each other in the plant. So yeah, this is quite common nowadays for physical AI for European companies also, yeah.Grace Shao (1:03:35)Mm-hmm.But in fact, actually you mentioned CUKA and it was bought out by Matee, right, a couple of years ago. So you’re also seeing a lot of Chinese companies like in the embodied AI, physical AI space actually actively buying out traditional brands in in Europe. How is that received actually locally?Alex Lu (1:03:47)Yes. actually for the for the embedded robots humanoid robots, there are not so many MA of Chinese players acquiring European companies. So basically I think for the humanoid robots, let’s say the robots like AJ Bot or like Uni3, China is much more advanced. And there was one robotic company in France, but they are kind of in financial difficulty. And another robotic companyThey were in they are invested by Renault in France, but still their technology if you look into that is not as advanced as Unitree or AJ Rob A Gi bots, for example.Grace Shao (1:04:42)I see. one last question is just do you think it’s fair that we’re overgeneralizing all the European companies into just one EU right now? Or do you think actually a lot of different com countries have different goals, ambitions, or even, you know, future tracks for them laid out?Alex Lu (1:05:03)very good question. So I can only when I see European companies, sorry, actually I’m thinking about French and German companies. So actually I cannot represent all the European countries and for different countries like Spa Spain, Italy. I’m I’m not familiar familiar with the country. I didn’t live there. I I didn’t receive enough clients from from these countries. So actually you are right.when I talk European companies, I’m more thinking about French and German companies. And of course, they are pr pretty different.Grace Shao (1:05:35)Okay. Well, thank you so much. Yeah, thank you. I just think it’s such a unique perspective because, you know, it it’s it’s more it’s easy for me to find someone who tells me the pure European perspective. It’s easy for me to find someone in the China US, but it’s harder for someone to for me to find someone f you know, who straddle between Europe and the Chinese market. You know, it’s obviously not as mainstream. So I’m really appreciative of your time and your insights and your sharing. Thank you so much, Alex.Alex Lu (1:06:04)Thanks,Grace. Yeah, thanks a lot again for i inviting me and accepting me for the podcast. And thanks a lot for your audience. And yeah, let’s keep in touch if any chance happens. we can have another talk if needed.Grace Shao (1:06:17)Definitely.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  China’s internet ecosystem, manufacturing base, batteries, EVs, robotics, and semiconductor becoming an AI-enabled industrial system 01.06.2026 50λ

  In this episode of Differentiated Understanding, I spoke with THE TP Huang, an independent China tech analyst known for his work on fintech, EVs, batteries, AI, semiconductors, and the broader China industrial ecosystem.The conversation traces China’s technology evolution from the early internet era to the present. TP argues that China’s internet ecosystem was shaped by a combination of censorship, protectionism, local engineering talent, and intense competition. That created powerful domestic champions such as Tencent, Alibaba, Huawei, Baidu, and ByteDance, which later became the foundation for super apps, payments, e-commerce, cloud infrastructure, and AI.The discussion then moves into China’s shift from software and internet platforms into hard tech: EVs, batteries, robotics, drones, semiconductor supply chains, and AI-enabled industrial systems. TP emphasizes that China’s technology companies are unusually willing to enter each other’s markets. Xiaomi moved from phones to chips and EVs; Huawei moved from telecom to semiconductors, AI chips, and autos; BYD moved from batteries to cars, solar, transit, chips, and potentially robotics.A major theme of the episode is that China’s AI story is not only about large language models. It is also about the physical stack around AI: batteries, sensors, motors, chips, power systems, critical minerals, factories, and real-world deployment. TP argues that this manufacturing and supply-chain density may become a major advantage in embodied AI and robotics, especially as real-world robot data becomes more valuable.Follow TP Huang here on X or Substack here To find the previous episodes of Differentiated Understanding, see here.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.Season two will host a series of guests from early-stage investing, as well as builders, researchers, founders, and product managers. For more information on the podcast series, see here.Chapters00:00 The Evolution of China’s Tech Landscape05:58 China’s Internet and Tech Sovereignty09:01 Investment Trends in China’s Tech Sector11:04 The Role of Government in AI Development20:00 The Intersection of EVs and Robotics26:07 China’s Competitive Edge in EVs and Robotics36:18 Global Strategies of Chinese EV Companies42:31 Advancements in AI and Robotics in China48:31 China’s Digital Infrastructure and AI Adoption57:38 Underappreciated Developments in China’s Tech Landscape01:00:00 Non-Consensus Views on China’s Economic HealthAI Generated Transcript (for reference only)Grace Shao (00:00)Hello everyone, welcome back to another episode of Differentiated Understanding. I am your host, Grace Shao. As many of you know, I also write the newsletter AI Proem, which is AI PROEM on Substack, so do give that a follow.Today we’re doing something special. We’re doing an audio-only version. I’m joined by TP Huang, an independent China tech analyst who writes about the intersection of fintech, EVs, batteries, AI, and broader China industrial policy. He has built a large following on X and Substack by combining data, supply-chain detail, and geopolitics to explain where China tech is actually heading.In this conversation, I want to use TP’s lens to understand the bigger China tech landscape: how China moved from internet platforms and payments into EVs, batteries, robotics, and now AI-enabled industrial systems. And since he quite literally said, “I can talk about anything China tech,” when I reached out, this conversation may follow the themes that I prepared, or really just go anywhere it naturally takes us. Very excited to have him on. Welcome, TP.Grace Shao (00:02)Hi, TP. Thank you so much for joining us today. I just did your intro before talking to you. And I told everyone that when I emailed you and reached out, I said, here are some topics I want to talk about. Is that okay? And you quite literally said, “We can talk about anything China tech.” So the conversation today could cover quite a lot of bases. I’m so excited to hear from you and have you kind of dissect a lot of your knowledge for us. And, you know, I’ve been a big fan of following your Twitter, your X, for a long time. Anyhow, thank you so much for joining us today.TP (00:33)I’m just really glad to be here, Grace.Grace Shao (00:37)Yeah. So you’re a mysterious man. Give us some color on your background and why you are so knowledgeable about China’s tech ecosystem, because you’ve really been covering everything from robotics to LLMs to the internet era. You cover them all, including hardware and chips and everything.TP (00:56)Yeah, so it’s kind of interesting that my actual background is not very technical in that area because I’ve been working mostly in the finance sector, or fintech sector slash crypto, for most of my working life. And I did spend a year recently working in an AI firm, so that was something different. But now I’m back to doing more crypto kind of stuff. So my background, I guess now, is a lot more AI-related.But a lot of the interest I had back in the day was in the renewable space and climate change and things like that. So that really got me started following solar panels, wind turbines, and then EVs. I first read about BYD back in 2008, like a lot of other people. And then as EVs were really taking off in China, that’s when I thought, okay, I really need to understand the full tech stack behind it. So that kind of got me into the entire battery supply chain, a lot of the upstream stuff, and then chips.The chips part became such a big deal because of AI. So then we had the October surprise back in 2022. That’s when I decided, okay, I’m really going to try to understand how the semiconductor manufacturing part of it works also. And thankfully, I was able to be connected to a lot of people. That allowed me to really understand a lot more.So I don’t profess to be an industry insider or anything like that. I’m just talking to other people who are working in the industry for some knowledge and writing about it. And then with AI, I actually worked on my own, no, not on my own. I worked with an AI startup, and one of the projects we did was actually for an AI toy. So I had experience running what I would consider to be AI robotics efforts. So I have a lot of real-time experience with embodied AI and also just using large language models. That’s kind of how I got into all this stuff in the first place.Grace Shao (03:26)It’s really cool because you have experience across the whole array. One personal question is: what drives you to really continue writing? Because you do write prolifically on Twitter. You have these hot takes, you put things together, and I think you’re quite widely followed by anyone who covers China tech. So what makes you want to share things publicly?TP (03:49)Yeah, I guess it’s more like a personality kind of thing, where I really just enjoy writing. And I think there’s something missing in the information space about what is going on in China.Last summer I was in China for a month, and I plan to be in China again for a month this summer, and I just saw a lot of really cool stuff. I think it’s good for the world as a whole to understand what’s going on in China, for Americans and for all Westerners to understand what’s going on in China, so that we are better informed in understanding how people can work with China and what kind of things people who want to compete against China need to know. But as a whole, I think it’s better to get proper information out there.And because China is a different language, and most people in China post in their own internet ecosystem on Weibo or WeChat, people don’t really read this stuff. So they get their sources from very bad sources on the English internet. A lot of them are just missing the nuance of what’s actually going on inside China. So because there is this vacuum, I just felt I’m obligated to actually do something about it, to help everyone understand better.Grace Shao (05:31)That’s awesome. It’s part of why I write AI Proem too. Well, okay, let’s get into the real stuff today. You’ve been following China’s tech for a while, like you said. Help us understand, just with the sentiment shift, how you view the early internet era to today’s success in hard tech and AI. What really has propelled China’s success in the tech sector in the last 10 to 20 years?TP (05:58)Yeah, so I think if we look back on things, China made a pretty big bet on developing its tech sovereignty back in the early 2000s and 2010s. It put a lot of policy in there under censorship reasons. It said, we’re blocking, we don’t want Google or whoever wants to enter China to actually censor the search results so that it fits our local law. And then what actually ended up happening was it became more of a protectionism kind of thing. So China was protecting the local tech champions at the same time that it was pouring a lot of money into these firms.So it allowed firms like Tencent and obviously Huawei and Alibaba to grow up. Later on, China also developed ByteDance. And if you look at how things are around the world, most countries, most leading Western countries that could have possibly developed their own tech ecosystem, like European countries or Japan, didn’t do it. The only other country that has a pretty robust local tech ecosystem or tech champion is Korea with Naver.And if you go to Korea, you notice that if you’re using Google Maps, it’s almost unusable. You kind of have to use Naver. So I think there’s a clear correlation between blocking US tech and some level of protectionism to having a local tech ecosystem being developed. And obviously it requires good local engineers also, so that they can take advantage of that. But China had all the ingredients for it.So even though it started maybe a decade after the US in developing this ecosystem, it was able to develop it because it didn’t have to face this immense competition from the US right away.And I think also there’s a lot of, you know, we talk about involution in China. I think there were stories of how when Uber tried to enter the Chinese market, because they had to face all these local Chinese companies that were working under 996-type hours, they were eventually pushed out of the market. So I think those are really the interesting parts of how the China tech scene developed in the 2010s.Grace Shao (09:01)Does that kind of feed into what we’re seeing now? Because right now it’s a completely different set of technology, yet in many ways it is building off the digital infrastructure that we just talked about, that got built out in the last 10 years or so.TP (09:17)Yeah. I think as a whole, if you go to China, even the internet ecosystem works entirely differently from America. In America, for the longest time, we had a search-oriented internet. You use Google, you use a lot of Google products, or you use social media. Whereas in China, because Baidu was never that great, people kind of advanced right away toward these mega apps like WeChat and Alipay.And as part of the movement on these fronts, you have these giant ecosystems developing where they not only have their own super apps, they also have their own e-commerce networks, their own payment systems, and they all got enough resources to eventually build their own cloud infrastructure and now develop into the AI world. So some of the biggest players in China when it comes to AI are the usual tech giants like Alibaba and ByteDance.Grace Shao (10:35)Yeah. So okay, let’s move on from that, from that big holistic overview of China’s internet space and tech sector. So much of the investor focus right now is still through the old internet platforms, like we mentioned, because of the natural progression of how they also become the major players in AI.But what kind of breakthroughs and capital moved from apps and payments into EVs, batteries, robotics, AI, and hardware? Are we seeing that these hyperscalers or big tech companies are also the major players in these other technologies that we’re talking about? Are they the main investors and backers, or is that a completely different ecosystem?TP (11:17)Yeah, China is kind of interesting to me in that a lot of the players are so uber-competitive that they are willing to get into other people’s spaces. So we saw Xiaomi move from the phone into developing their own pretty advanced AI team. They have their own chip design, and most notably they have their own EV division, which is doing really well.We saw Huawei start off in telecom and then move into the entire semiconductor ecosystem, and also their AI chips, and also into the auto division. We saw BYD start off as this battery company, and then it got into all these areas. It got into cars, it got into solar panels, it got into public transit, it got into the chipmaking side of things, and now it’s also looking to get into robotics with humanoid robots.Whereas you don’t really see that as much in America, where it’s mostly a typical thing I used to listen to on Wall Street, this entire idea of capacity discipline. Which is basically: how do we reduce competition so that we can get a higher margin? Whereas the Chinese marketplace seems to be one where everyone’s trying to squeeze in at the same time and just fight it out until whoever has the best cost controls ends up winning.From that point of view, I think this is why for some time people saw that the Chinese stock market hadn’t been growing as much as the US stock market, because there’s just so much competition inside China. So a lot of the funding for these efforts inside China actually had to be backed by the government, these big funds and things like that. And also these things, they are willing to put money into areas of lower initial returns.A lot of the car factories, maybe they’re not the best investment if you’re looking for a 100% return. Maybe it’s not the best for that. But because it provides local jobs and things like that, the government is willing to put some money into it. And we saw that right now with semiconductors also, and also the data center build-outs. So that is how, over time, the entire Chinese manufacturing ecosystem kind of got built out.America is trying to do a little bit of that right now with AI data centers and trying to do that with the tariff wars. But fundamentally, the market in the US is about squeezing out competition and lowering capacity in order to charge more. Whereas the Chinese system is about how to scale up production and lower the cost of production in order to have higher margins. So it kind of works differently.Grace Shao (14:43)Yeah. So on top of government help and actually putting money into sectors that often have lower initial returns, sectors that are not so sexy in the beginning, let’s talk about DeepSeek.I think it’s been interesting because we know DeepSeek and many of the other Chinese labs weren’t getting a lot of capital until maybe 2022 or 2023. However, now they’re obviously being pushed front and center as the main economic drivers. Not only are they being looked at as very sexy investments from the private side, but the government funds are also looking to put cash behind this.How do you view the relationship between government policy, government mandate, and the AI labs in China? That’s part one of the question. Part two is, if DeepSeek and a lot of these Chinese labs permanently price their models at, say, one-thirtieth of the American labs’ prices, what’s the thinking on that? And what’s the sustainable business model for them looking forward?TP (15:44)Yeah, so I think it took a while for China to really catch on to this entire large language model thing, because a lot of the Chinese AI, when I looked at it back in the early 2020s, was aimed at embodied AI. So in terms of smart manufacturing, how to improve the grids, drones, robotics, and also EVs, things like that.Whereas a lot of the US funding for AI was, I guess, kind of abstract. You want to develop the best models, and then we will find the use cases for them. But once it took off, I think there was kind of a light-bulb switch inside the Chinese sector that we can’t just let this go, we have to catch up. The way Chinese people think about things is like, we have to get in on these opportunities.So in the beginning, with Chinese large language model development, I think it was mostly the big tech companies like Baidu that were kind of leading the efforts. But over time, more recently, I think you find that it’s the startups that have done some unique research that have done the best, like DeepSeek, obviously Kimi, and Z.AI.And obviously some of the big tech companies are still quite successful, like ByteDance. They have a very good AI product. And Alibaba, with the Qwen stuff, is also very well developed. But you do see that the Chinese government, ever since the DeepSeek moment, has been investing more in funding to make sure that the domestic AI startups are able to get the funding they need to compete.In the most recent example, DeepSeek, they actually got paired up with Huawei, or maybe they came together somehow. But you can see just in the V4 release recently that there was a lot of integration work between Huawei and DeepSeek. The DeepSeek models are deeply integrated, so that you can use the Ascend chips from Huawei to better train and run the models. And this is part of China’s overall strategy of being self-sufficient in both the hardware and software side of things for AI.So even though it’s probably easier to just buy NVIDIA chips, the risk of getting cut off by the US government is pretty high. So it’s in China’s long-term interest to have its own ecosystem across the board.No other country has that. China has not only the chips and the software, but also the entire AI data center build-out ecosystem. There has been a lot of investment or money put into AI build-out-related stocks recently, like optical modules, optical transceiver suppliers, fiber cable suppliers, PCBs, power chips, and things like that.So there is a lot of investment across China, not just in the software part of it, but also in the hardware integration part of it. And at the end of it, it’s all supported by the Chinese government in some way because they want to make sure that they have a domestic supply chain, so they can’t just get cut off at any point.Grace Shao (20:42)So you’re basically in the camp of what Jensen was saying: export controls are not working. In effect, they are cutting American suppliers or vendors out of China, and in that case, actually pushing China to become more and more self-sufficient.TP (20:57)Yeah. I mean, for a long time there, Jensen and the good people behind SMCI were trying to get as many NVIDIA chips to China through backdoors, or through Asian and Southeast Asian data centers, as they could, right? So that the Chinese AI suppliers remain hooked onto the NVIDIA ecosystem. But you can see that by sometime late last year, the Chinese government was actively blocking these things from happening because they really wanted the domestic AI players to use the local ecosystem.Grace Shao (21:39)But is it actually being replaced right now? Or do you think in the short term, medium term, long term kind of thing? The long-term strategy is self-sufficiency. Short term, it doesn’t seem like it’s realistic yet, right?TP (21:51)Yeah, so this is the interesting part. For much of 2023 and 2024, what the Chinese players were doing was that a lot of them were importing the permitted versions, like H800 and H20s from NVIDIA, through official channels.And then there was a lot of smuggling of chips into China at the same time, and the Chinese government was allowing this. So whenever they were building AI data centers, they would have the data centers that use domestic chips and ones that don’t use domestic chips.So what would happen is, let’s say Alibaba was looking to access NVIDIA compute and it doesn’t want to get sanctioned by the US government. So what it would do is, it buys some NVIDIA H20s, puts them in a data center, and also leases compute that runs on NVIDIA from one of the state-built or local government-built data centers that smuggled in chips, because it didn’t want to get in trouble by buying them if it’s not allowed to.Another thing that these firms started doing that’s entirely illegal, again, is actually just setting up companies offshore that would buy these NVIDIA chips and then build data centers in the rest of Asia, places like Japan, Thailand, and Malaysia. And then they would lease the compute for these NVIDIA chips from these data centers.And that’s still going on right now. The Chinese government is allowing that because domestic firms like ByteDance would just say to the Chinese government, we need this ability to use American chips in order to not be left behind. Because if you talk to the AI developers in China, they don’t enjoy using Ascend libraries for training. They don’t mind using them to run inference, but for training, they still prefer to use NVIDIA chips.So there is an effort right now to also get the training part of it up to par. And that’s kind of what the DeepSeek work with the Huawei team in recent months has been about. It’s kind of interesting to see how much better the integration has made the Ascend chips run training and inference on the DeepSeek models.There has also recently been a Qwen model called 3.7 that came out. And they also released their own AI chip called Chengwu MA90.TP (25:10)And part of the interesting thing about that is not only did Alibaba have the self-designed chip, because it was designed internally, it used its own internal AI models to write the kernels for the chip. And it had some really good results. I think going forward, a lot more of these domestic chips will actually be able to at least do part of the training also.Grace Shao (25:39)That’s really interesting. And it actually echoes some of the stuff I’ve heard on the ground as well. So like I said in the beginning of our conversation, I don’t want today’s conversation to only focus on China’s LLM and model space. I want to double-click on something you mentioned at the very beginning of this answer. You said China actually started with its capital focus and technological focus on EVs and embodied AI.What’s interesting is that that side of things didn’t really pick up in the US or in the West, per se, until more recently. So did the EVs come first, or did robotics come first? Or did they kind of converge and come at the same time, and there’s synergy there?TP (26:23)Yeah, so when it comes to the EV and robotics story, I tend to think of it as something that started because China was doing all the manufacturing of consumer electronics. And that’s how it was able to then develop these OEMs in the smartphone space, like Xiaomi, Huawei, Vivo, Oppo, and Honor. Basically, they developed this entire workforce inside China that was very good at dealing with supply chains and also integrating things together and doing manufacturing.I personally had an experience with this about a year ago, where we were trying to make this AI toy, and I got on a call with a Chinese factory. I won’t say which one. But basically, about five minutes into it, I realized America was in trouble because we had all these American engineers who are decently smart people. And the sales lady at the Chinese factory just knew way more about how hardware works and should work than any of us did.It was a very humbling experience just to see how there’s a lot of process knowledge involved in this. There’s a lot of experience involved in this stuff, right? And my cousin actually works in Shenzhen.Grace Shao (27:46)It was like learning from experience instead of PhDs, right?TP (28:03)They developed their own automated device that tests blood samples to see what kind of disease you might have, something like that. And what I realized talking to him was that this entire supply chain in China around Shenzhen or around Hangzhou is very deep.Because of that, a lot of the modern tech that we see with embodied AI comes from this basic understanding of supply chain, software-hardware integration, and also electrical platforms. What are the commonalities between drones, robotics, cars, and EVs, right?First, you need to have this battery underneath. You need to have electrical platforms. You need to have PCBs. You need to have cooling systems involved. You need to have control chips. You need to have power management chips. You need to have main control chips for the actual device. You need to have AI chips.All this stuff, in the beginning, Chinese suppliers were sourcing from abroad. Over time, due to export controls, they started doing this domestic substitution. They’re still the biggest importer of chips globally, but a lot of that stuff is coming in-house now.So if you do a teardown of a DJI drone, you’ll probably find memory chips from CXMT and YMTC. You’ll probably find CMOS chips for the camera modules from maybe OmniVision or something like that. And the battery is obviously going to be domestic. And all the stuff that we saw with drones and with EVs, we’re now seeing with humanoid robots and other kinds of robots, because at the end of it, a lot of the basic concept is very similar.You need to have some kind of a brain for the embodied AI machinery. And then it needs to have some kind of battery source to actually do the functionalities. And then it needs to move using some kind of motors, and then it needs to be able to absorb information from its surroundings with these sensors.That is why China has such a large ecosystem, because it has a good upstream supplier network and a lot of people working on this stuff. Whereas if you come to America, there’s just not a lot of that talent around.So if you want to develop an AI robot, you have to do everything in-house and figure it out. Because if you can imagine, if you don’t develop in-house and you contact a supplier in China, you can’t really iterate things quickly because you’re working with someone over there who doesn’t speak English and also doesn’t work the same hours you do. So the turnaround time is just much slower.Whereas if you have an idea in China for an AI robot that you want to build and sell to the market, you can get it produced in a month. That would be crazy for any kind of AI startup in America to do.Grace Shao (31:59)Yeah. In fact, I think there are a lot of robotics companies right now with founders who are literally tweeting about this thing: we must move to Shenzhen. Or I know of companies that actually get their hardware completely end-to-end, basically buying from OEMs from Shenzhen and slapping on a tag elsewhere.But I want to ask, why did China ultimately come out on top in EVs? Because from what you just mentioned, technically wouldn’t countries like South Korea have an edge? They have car manufacturers, they have chips, they have memory chips, especially when you just talked about brains. It’s not like the brains that we’re talking about right now are AI brains yet.So what made China actually come out on top with EVs and robots? Was it again this narrative around government push, because the country needs clean air? Was it because of innovation? Was it because of renewables and everything coming together? How do we understand this?TP (33:01)Well, I think Korea itself is actually a country with a lot of industrial policy also. So I wouldn’t necessarily say that the Koreans were less aggressive about government support than the Chinese were.I would say that if you look at just the human capital side of things, we’re looking at a magnitude difference in the number of engineers coming out of South Korea and China. So that’s something not easily made up.If you have 10,000 battery engineers from China every year, and let’s say you have 1,000 from Korea, the 10,000 are going to crush the 1,000 over time. And you can kind of see that. Back in the late 2010s, the Koreans were ahead of China in battery technology. But because Chinese industries were moving so fast and the supply chain was moving so fast, China has been ahead of Korean battery makers for several years now. And the gap is only expanding as we move toward more advanced solid-state batteries, or lower-cost sodium-ion batteries.Batteries are such an important part of the modern electrical transition that it’s kind of mind-boggling that China controls so much of the entire ecosystem. People keep talking about TSMC, or Taiwan having some percentage of manufacturing for chips, which by the way is not true. But Taiwan only has a small part of the entire ecosystem. Korea only has a small part of the semiconductor ecosystem, right? America has a huge percentage of the semiconductor ecosystem.But if you look at things like rare earths, critical minerals, and batteries, China actually probably controls 80% to 90% of these ecosystems. So even the Korean battery makers rely on the Chinese supply chain for a lot of their inputs now. And there’s just no way to get around it because the Chinese process knowledge, cost advantage, and engineering advantage are very hard for a smaller country like Korea to overcome.Grace Shao (35:47)Interesting. Yeah. So how should we understand these companies’ international strategies? Because I think you’ve written about it before. Like you said, they are major exporters. How do the battery companies and EV companies position themselves globally? Are they quite aggressive? Are they suppliers along the supply chain? Are they building up consumer brands? How do we understand that?TP (36:19)Well, it’s different with different people. I think because the domestic market is so aggressive and so competitive, companies like BYD had to go abroad to get higher margins on their products. That’s kind of forced a strategy where they’ve aggressively expanded. Things especially picked up in the past few months because of the Iran war, where there’s also a lot of demand for these EV products abroad.And as a result of that, it helps what I call China Inc. As you see more of these high-tech EVs abroad, as you see more of these DJI drones and Chinese AI models abroad, there is a generally higher view of Chinese products now from much of the Global South. And as a result of that, Chinese firms are also having greater success selling their products.I think one of the interesting things recently is just to see how much the Chinese automakers’ market share in Europe has already surpassed the Koreans and is catching up to the Japanese. Just looking at that, it gives me the impression that the Chinese automakers, and just China Inc. as a whole, have gained a reputation for quality in a very short period of time. And you can only do that if the automakers themselves are making a real effort to build their brands and promote their products in these markets.And I think they’re getting paid off because my guess is that BYD’s automotive sales have much higher margins on stuff sold outside China than inside China.Grace Shao (38:43)I see. So it’s still like a pricing strategy, or winning on pricing, you’re saying.TP (38:50)I think in China it’s more of a pricing strategy, but abroad you see them actually marking things pretty high. So maybe there is a pricing part of it, but if you listen to Stella Li, Executive Vice President of BYD and President of BYD Americas, talk about the new models that they launched in Europe, they’re very much trying to frame it as a luxury brand, with the Denza model brands.She would say that this is technology that does not have any competitor or equal in Europe. We’re just way ahead of the Europeans here. We’re going to build the fastest charging network that you’ve ever seen. You can charge your car in five minutes, for example.It’s kind of interesting because BYD can sell its cars at a much higher price outside China than inside China. Inside China, it might have to sell its cars at a discount to Tesla cars. Outside China, it might sell them at the same price as a Tesla car. So yeah, I find that interesting.Grace Shao (40:04)That’s very interesting. And I’m kind of playing devil’s advocate purposely. Anecdotally, I’ve obviously been in a lot of BYD cars when traveling in China. They are actually really, really sleekly designed. And like you said, in China, for some reason, they’re positioned more as not a luxury car at all.But even in Hong Kong, I’m seeing more and more Zeekr cars and BYD cars taking the roads, and they’re definitely replacing previous Audi and Volvo owners. It’s very interesting that that’s the trend. Outside of mainland China, the reputation of these Chinese EVs is almost more premium than they are in China.TP (40:49)Yeah. And one of the reasons BYD wanted to do well in Japan and Germany was that it thought that once it started selling well in Japan and Germany and got approved by those automotive nations, people inside China, especially suburbanites in Shanghai, would then accept BYD as quality products. It is kind of interesting that a lot of times the Chinese can’t really accept that we have quality products unless it’s also being accepted abroad. It is kind of interesting how that works.Grace Shao (41:21)Psychology, I guess.TP (41:31)Yeah.Grace Shao (41:37)I guess it’s a little bit of a psychological play on this as well. I do like your framing on China Inc. And I think recently we’ve seen that even in the consumer space. It was so interesting that Luckin Coffee bought Blue Bottle coffee, and you’re getting more and more of these kinds of purchases, like SHEIN buying out Everlane, etc.But I want to bring it back. I want to bring it back to AI.You said earlier that China’s mastery of hardware manufacturing has given it an edge in scaling humanoid and service robots. But how do we understand where we are with world models and the actual next stage of embodied AI and physical AI right now? Because like what we just discussed, China’s manufacturers are very experienced in building out the robots, drones, and various forms of robot mechanics. But where are we with actually injecting that with AI?TP (43:02)Yeah, so I’ve been in touch with the guys behind the China Research Collective, and they are actually inside China, so I’ve had some discussions with them about this. They’re telling me that because China has this hyper-competitive local market for jobs, a lot of young people are having trouble getting the jobs they wanted. So they’re willing to help these AI companies collect data on doing things to help these world models.It’s kind of interesting because you need a certain amount of data so that the robots can simulate human movement and then do the tasks. But at a certain point, if you have a child, you know that it takes them a long time to be able to walk around and then run, because they need to first feel and touch everything and learn everything over a year or so. During this time, their muscles develop and their muscle memory develops so that at a certain point they no longer need to think about how they walk. They can just walk. They no longer need to think about what they can or cannot eat, because they already put that stuff in their mouths to test it out.Longer term, I think once you have enough robots in China, they will just be able to improve exponentially in their capabilities because they will be able to fast-track all this, what I call reinforcement learning in the real world. If you try grabbing an object a million times, eventually you’ll figure out the best way to grab it. And once a robot learns how to grab it, that gets shared amongst all the robots of that family.So I think as you see the Chinese robotics rollout speed up, this is when you see this decisive edge in the world models. We already saw this with drones, right? The Chinese drones are just so much better at moving around and doing stuff because they had so much more data than anyone else.We’re seeing it now in EVs, where the Chinese self-driving cars are really good because they’ve had a lot of data out there, where people are just using autonomous features to do all the work. And you’re seeing that BYD today is having this entire unveiling where it’s talking about its path toward L3 and L4 autonomous driving.The more data it has, the better it’s going to get. That data becomes an advantage going forward. In the future, whoever has the most robots out there in the real world, and has all that data, can then train their robots faster. That’s why it’s kind of a big deal right now that BYD says it’s going to have 20,000 robots in its factories this year, because then it has all this data on using robots in a factory setting. That’s going to improve the performance of the world models by leaps and bounds.Grace Shao (46:48)Mm-hmm. Because the biggest bottleneck right now is just not having enough 3D data. And collecting that kind of 3D data is extremely challenging without, like you said, real, actual physical deployment. That’s fascinating.TP (47:10)Yeah. I also want to point out one other big difference between the Chinese players and the foreign players outside China, which is that China has this entire critical mineral supply chain. That is foundational to the rare earth magnets, for example, needed for the different robots and EVs, and for the motors, and also the materials needed to build the humanoid robots themselves, like magnesium. It produces about 80% of the world’s magnesium, and magnesium alloy is considered to be the main material that you want to use for humanoid robots.Grace Shao (47:57)I want to tie it back to what we also talked about earlier. Does the very strong digital infrastructure layer, just from fintech, IoT, and 5G, now contribute to China’s very quick adoption and diffusion of AI in the real economy? And how do you view this kind of positive cycle versus in other economies, where sometimes the digital infrastructure maybe just isn’t there yet and seems to need time to build up as well?TP (48:31)Yeah, I actually think this is one area where America might have a leg up on China, because the American big tech companies tend to also be the biggest cloud service providers. The Chinese ones are a little smaller. So right now, you only see the competition between the US and China because they’re the only two countries that have this data center and AI infrastructure advantage over the rest of the world.The biggest players in China, like ByteDance with their entire AI cloud infrastructure and their entire AI app ecosystem, are also the ones that are able to deploy their apps globally the fastest. In America, ChatGPT/OpenAI has this commanding position not because it has an ecosystem, but just because it was the first to do it. It had a first-mover advantage.But if you look at the players outside of ChatGPT, it’s Google slash Gemini that probably has the largest market share, because it has this big data center hardware, this AI infrastructure advantage over other players. And also it has this app system that people can use the AI features in.In China right now, personally, I don’t get to use the AI apps in China all that much, but I do have a Chinese phone, and I use ByteDance’s Doubao app, and it’s really good. So that has allowed ByteDance to have the best video generation model out there, called Seedance 2.0.Grace Shao (50:29)Mm-hmm. And they really leverage and lean into their data advantage as well. Obviously, if you own TikTok and Douyin, you have the most amount of video data in the world.TP (50:43)And not just that, they also have CapCut.Grace Shao (50:58)They do, which is the editing tool. I actually use it to edit our videos here on AI Proem. It’s great. I kind of want to wrap it up soon.I want to ask you a forward-looking question. If we connect the dots from your fintech days covering the digital economy to what we just touched on, EVs, robotics, hardware, everything, where do you think China’s digital economy goes over the next five to 10 years? What are the biggest bottlenecks? Will that look very different from the rest of the world? Or do you think the evolution of technology will be organic and go in the same direction, no matter your geographical location or your domestic strengths or weaknesses?TP (51:30)Yeah, so I will first talk about where I think they can possibly see the most improvement, and that will be the semiconductor part of it. I do think they will have a fully domestic semiconductor supply chain pretty soon. And that, along with government support in terms of putting money into these high-capex, maybe lower-rate-of-return investments, will allow them to more aggressively build out the domestic semiconductor infrastructure.Once you have that infrastructure, then you can produce all the AI chips, all the phone chips, and all the analog chips that you need for your various embodied AI products and EVs and all these other leading sectors. And once you have that, that means you’re no longer constrained. You’re no longer constrained by compute. You’re no longer constrained by possible Western tech export controls on you.So then the AI players in China are equal in terms of AI infrastructure. And that allows them to compete a little bit better with their American counterparts. Now, they do have some obvious advantages over their American counterparts. We’ll have to see how this plays out, because China does have this entire grid build-out that is just unrivaled. And as we move to a more electrified global economy, being able to build not only data centers but the entire grid is actually a huge competitive advantage over the rest of the world.I don’t really want to say who wins the AI race, because I feel like you can only lose the AI race by not participating and investing in it. But if you invest and put a lot of money into it, like both the US and China have, both of these countries will have a huge share of the global economy going forward.TP (54:15)I just don’t see how you can put this much effort into AI in America and not get something out of it.Grace Shao (54:24)I just feel like it’s not a zero-sum game.TP (54:28)It’s only bad if you don’t try to build your own AI industry, right? If you don’t invest, that’s a problem. But if you invest, something good will happen, I think.Grace Shao (54:42)What about the smaller countries where they don’t have that capital, and maybe they don’t have that much capital to deploy into this, or even frankly the talent to build their whole AI stack? Where do they fit into all this?TP (54:53)Yeah, so I think that’s one of the factors that might help the Chinese ecosystem over time, because a lot of the open-source stuff is coming out of China right now. So if you’re from one of the smaller countries, let’s say Singapore, and you want to develop your AI sector, you are more likely to use an existing open-source model and do reinforcement learning training on top of that, and then develop your AI product on top of that, than use something you don’t have any control over, like Claude, for example.Grace Shao (55:41)Interesting that you use Singapore, because I was just there last week and literally OpenAI just announced their satellite office. I think they said they would employ 200 people. Singapore is an interesting story because, if anything, they’re super gung-ho on AI, from top-level diplomats and ministers to companies. So it will be interesting to see how they play out this strategy.My question for Singapore is: they can attract a lot of talent globally to go over. They can attract a lot of new companies to go over, which is what they did with the internet era too. ByteDance, Tencent, Facebook, everyone’s there. But then what is the value they propose for the locals? Or how do they plan to diffuse AI into the economy? I don’t know how they make themselves that relevant globally beyond being a hub for these companies.TP (56:35)That’s a very hard thing to say because I don’t see Singapore, just on its own population, actually developing anything unique. The people who would work in Singapore’s AI industry could work in any other country also. So I think Singapore has always put itself out there by being a country that attracts talent from all over Asia, right? And they attract a lot of capital also from the rest of Asia.There have been a lot of issues in recent years where they say all this money coming in hasn’t really helped the local-born population in Singapore. So that is something interesting to watch out for.Grace Shao (57:25)Yeah. I don’t want to go on a tangent on Singapore too much. So, last two questions. One is: what is one underappreciated hard-tech development you think people are missing?TP (57:38)Yeah. Last year, I wrote a thread about a list of what I call sanction-breaking tech that was happening in China. A lot of these are not things you see in the media as much, because they are the zero-to-one steps in the upstream supply chain that need to be achieved in order for an end product to be built three or four years later.So things like high-speed analog-to-digital converters and digital-to-analog converters, advanced diamond substrate for heat sinks and other purposes, high-end gallium chip designs, and a lot of the lower-level material science-related stuff that people don’t really see.But once China develops these things, that’s when you see this really fast iteration afterward. Because everything in China is kind of built upon the idea of having the upstream supply chain and the process knowledge. And then it can iterate through the end product a lot faster.So as fast as China has moved in the past 20 years, I don’t think the West is really prepared for what is to come out of China in the next 10 years. I really don’t.Grace Shao (59:36)Interesting. Okay. Well, I think that’s a topic that no one really has an answer to. No one really knows the future, right? But I appreciate your thoughtful answer.My last question for you is a question I ask everyone who comes on the show. What is one differentiated view you hold that you think is non-consensus?TP (1:00:00)Interesting. Well, one thing that I’ve talked a lot about with people recently is that if you listen to mainstream media, when they talk about China, they always talk about the economy not doing well and that China has this housing bubble that’s apparently a real problem, right? And that China has this demographic problem going forward, and that’s why China might have problems going forward.I’ve actually always held the opposite belief, in that I’m always under the impression that China grew overly rapidly for many years because it built up this real estate bubble, and all that money went to real estate instead of the tech sectors. And at a certain point, it decided that it could no longer blow up this real estate bubble because young people weren’t getting married and having kids because they couldn’t afford homes. So it deliberately deflated the real estate bubble in order to solve this problem.And then it still claims to have grown at around 5% a year for the past few years. If you can deflate a bubble and grow at 5% a year, that is quite the accomplishment, actually. So I would say the Chinese economy is quite healthy.You would rather have an economy that can grow strongly in the middle of an asset bubble deflation versus an economy that is growing just a little bit in the middle of a historically large asset bubble, like you have in the equity market in the US.Grace Shao (1:02:05)That’s a very interesting take, actually. I’ve never heard someone say that. But yeah, I kind of see where you’re coming from.TP (1:02:14)Yeah, that is my take.Grace Shao (1:02:17)I love it. TP, look, I’ve taken up an hour of your time. I really appreciate your insights. And you entertained my brain going in all directions as well. We’ve really talked about a lot of different topics today.Is there anything else you think we didn’t cover that you would like to share with everyone? Or do you think we can always pick this up again another time?TP (1:02:40)The only thing I would say to everyone out there is, if you enjoy AI, try one of the cheap Chinese models and see how it works for you. I’ve tried it myself. It’s great for my work purposes. And I highly recommend everyone use Kimi.Grace Shao (1:02:58)There’s a plug. No, I’m kidding. They are good, actually. I think I use different models for different things, but ultimately I find that if you’re really using them for more basic writing and everything, the Western ones are better. But if you’re really hosting your own models and running your own agents, then a lot of the Chinese ones are a lot more cost-efficient.So thanks again, TP. Thank you so much for your time.TP (1:03:28)I’m glad to be here. I’m glad to be on your show. And you can all follow me on X at TP Huang. I’m really glad to be on this show.Grace Shao (1:03:38)Definitely. And TP is on Substack too.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  China's open-source ecosytem and the future of AI bootstrapping with ex-Hugging Face APAC head 25.05.2026 1ώ 3λ

  Joining me today is Tiezhen Wang (Tom), formerly of Hugging Face, where he worked with researchers in China, Australia, South Korea, Japan and across APAC, to help make open-source models more discoverable, usable, and visible to the global developer community. In this conversation, Tiezhen explains why Hugging Face became the GitHub for models and why open source is not just a distribution mechanism but a different way of coordinating research. We discuss why Chinese AI labs have leaned so aggressively into open models, how DeepSeek changed the commercial logic of open source, and why Qwen, Kimi, GLM, MiniMax, and others are using openness as a way to win attention, recruit talent, and accelerate the whole ecosystem.His core argument is that China’s open-source AI push has three layers. At the researcher level, open source preserves attribution and career mobility. At the company level, open models can become benchmark-led marketing, developer distribution, and a recruiting advantage. At the ecosystem level, government and university incentives are beginning to cultivate open-source culture among younger engineers.We also discuss why US frontier labs have pulled back from openness as research and business have become more tightly coupled, why distillation is much murkier than the public debate suggests, and how DeepSeek’s releases increasingly function as shared R&D for the broader AI ecosystem. The conversation then turns to monetization: why open-weight labs can still make money through API tokens, base-model access, post-training services, and inference optimization.Finally, he lays out his current thinking on AI bootstrapping: the idea that agents may eventually help improve their own harnesses, generate training data, and even improve the models they rely on. We close on a more philosophical question: if a handful of closed labs control access to frontier capability, open source becomes more than a technical preference. It becomes a check on the concentration of power.Tiezhen/ Tom is based in Sydney, Australia. Feel free to reach out to him on X to chat.To find the previous episodes of Differentiated Understanding, see here.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently. Season two will host a series of guests from early-stage investing, as well as builders, researchers, founders, and product managers. For more information on the podcast series, see here.Chapters04:07 The Philosophy of Open Source at Hugging Face12:51 Challenges and Opportunities in Open Source17:12 The Role of Collaboration in Research21:50 The Future of Open Source and AI33:58 What Constitutes Distillation in AI37:18 Navigating Copyright and AI Distillation37:43 The APAC AI Landscape: Insights Beyond China43:08 Understanding the Ecosystem: Labs vs. Hyperscalers46:21 Monetizing Open Source AI Models52:02 The Future of AI: Bootstrapping and Self-EvolutionTranscript (AI- generated for reference only)Grace Shao (00:00)Tie Zhen thank you so much for joining us today. I’m really excited to have you on. We’ve been trying to make this happen for a while and just so glad the timing’s finally worked out. To start, can you tell us a bit about yourself, your journey, and where you’re at right now in your career and how you see the whole ecosystem? And also, just help us understand Hugging Face a little bit as well.Tiezhen Wang (00:19)Yeah, thanks, Grace, for inviting me. I know, sorry for the long delay. It has been a while, but I’m recently in transition because I just left Hugging Face. So to give you a quick information about very high-level overview, you can think of Hugging Face as the GitHub for AI. If you are not familiar with GitHub, you can think of Hugging Face as Amazon, where you can find all kinds of models in one store.And we are helping, so my job is to help researchers to get their models, which is the open source models on Hugging Face. And they can use the best, like all the tools, all the services on Hugging Face to make their models more discoverable and available to everyone. We also offer all kinds of technologies. For example, we allow them to create demos so that developers do not need to download the whole models.and they were able to try it out and see how it goes. And we also offer services so you can create your own agent using open source models. We do all kinds of scaffolding on top of open source models. another part of work that we do is to help them get more traction. We use LinkedIn.I use Twitter mostly to help them getting well known by the public. And we write analysis on their models and letting people know what are the new inventions from the model, et cetera. we work with researchers across the world. Like myself, it’s focused on APAC, especially Chinese researchers. Yeah, that’s pretty much the goal.quick overview of what I do. If you have any questions, just let me know.Grace Shao (02:03)And how did you get to this role? Because I understand you were with Google for quite a while as well.Tiezhen Wang (02:07)Yes, I was with Google as an engineer. work on ML frameworks. But then we had a bunch of reorg. And I was assigned to a project which is not open-source. But I really like talking to people in the open source world. It’s kind of very different. So when you are paid to work something versus you want to work on something yourself,Like you have very different mentality and very different feelings. So when I was working on the open source machine learning framework, I talked to people outside Google. And I can see the stars in their eyes. They do want to work on something they want. And even though they may not get paid, et cetera, I really like this feeling. So after I was assigned to the non-open-source project, I want to try something likenew but also in open source and I was like talking to people in Hugging Face and I really liked them. At that time, like Hugging Face was not like part of the mainstream. It was like a niche product for researchers where researchers can upload models. But I do see there’s a huge potential for Hugging Face to grow up because first I believe in open source and the second like Hugging Face is going to be the entry point where like all people will come in and search for open source models. But the most important of all is that I feel that Hugging Face is a company who understands how open source works. Open source is a huge leverage. If you use it well, it’s going to be very powerful. And Hugging Face is like 200 people, like very small companies compared to other companies growing up from the same area. But they are able to use open source as a leverage.and called for collaborations across the world and do very impactful things. a lot of people, a lot of big companies are doing open source, but they just don’t understand this age. That’s the essence of open source. And I do feel that Hugging Face is doing really well there. That’s one of the reasons why I want to join Hugging Face.Grace Shao (04:06)Yeah, I think that’s amazing. I think that’s something we definitely will double click on later, especially when we talk about why China’s labs seem to have been embracing open source. Just kind of one last question on just the whole ecosystem and how hugging face fit into it. What was the philosophy really held by the whole company? Because I actually listened to one of the founders interviews, Clem’s interview recently. And during the interview, he talked about how Chinese scientists have always been long term contributors to open source technology. And then he said it was really like kind of a pivotal moment around 2022 where American open source contributors kind of took a step back and then there was a sentimental shift in the ecosystem. Why is that and how does Hugging Face kind of view the whole ecosystem?Tiezhen Wang (04:47)Yeah, there are several questions. Let me try to address them one by one. The first one is the philosophy behind Hugging Face. I think it’s really the mindset. so anything that we see where we can have a collaboration, like Hugging Face will just reach out and see if we can collaborate. So if you go to see a lot of work released by researchers, they will have paper on arXiv.and also their project on GitHub. And you’ll see me on all of these issue number one, which is the first issue after the repository has been released. And we just write something saying, offer blah, blah, blah. Do you want to collaborate on something? So for anything that we can collaborate on, we will just call for collaboration. And some we’ll go through, some we’ll not. But this collaborative mindset is very, very different from.like a business point of view. From a business point of view, you will first think, what is my edge and how I win the market, how I compete with others, and what are the end areas. After the competition, what’s the end game, how it will go. So that’s the way of how you can justify the investment and everything. In open source world, it’s totally different. It’s like, I want to do something.I just say it and I do it and there are developers who want to join in and we do it together and we grow the pie gradually. we do not have like, let me put it the other way. So if you see an open source model coming from one of the Chinese lab, for example, GLM 5.1 is released and you may think like Kimi or Minimax like other open source model provider.in China would compete with them. But actually not. Like you will see they are commenting on the Twitter saying, congratulations, et cetera. This is a collaborative mindset where everyone is stepping up on each other. we can do a lot of, as a group, can continue to push the frontier forward. So I think this is very, very different.Yeah, and talking about your second question, the Chinese, well, I wouldn’t say labs. Chinese researchers, labs, companies, et cetera, they all want open source. I think there are three different folds. The first one is on the researcher side. A researcher would always prefer if their work is open source. That’s coming from their academia background, because when youLike on the CS world, when you write a paper, you have to show that it’s actually working. You have to show that all the numbers are real. Other people should be able to verify that. And you can only do that by releasing your code, releasing your models to the community so that other people can evaluate. So a researcher, after they graduate and they go to a company, they will bring this mindset forward. And by default, they are open source people.And another perspective is for their self, for the career development of themselves. So as an engineer in big companies, it’s very often that you are working on some project and nobody knows that you are working on that project until you say that out on the game or on your resume. But open source is very different. We know precisely who has contributed to DeepSeek before.And that’s very attractive for for researchers, because if I have done great work, I want the whole world to know that I’m doing excellent work. This will help me have better branding, help me to do more collaboration, help me in the future step in the career. So a researcher would always love open source, by default. So that’s the first part from a researcher’s level. The second one is from business level.So well for individual is quite easy to embrace open source from manager level from the executive, they need to justify the investment on open source. I have to spend tens of millions in training a model and you want me to give it for free. That’s crazy, right? That’s how people think before DeepSeek. Although we have lot of open source models before DeepSeek, but the trend is completely changed.Before DeepSeek, people were thinking, oh, maybe the model is not that good. Maybe I’ll just open source it. But if the model is good enough, maybe I’ll keep it for private. And that’s one of the reasons why you see a lot of people were saying open source is not that good, especially from Robyn. And lot of people do not understand how the open source works.works. But then people do realize that if they do not open source, they do not even have a chance to stand on the market. Because their model first is not really good. If they just compete on the marketing level, on the business level, they do not stand a chance, not even a chance. So you spend tens of millions and you get nothing. But if you open source, at least you have some sharing and people will remember. And also you can have the market from.for the researchers. I think Qwen team was one of the first team who understand it from a business level and start like open sourcing work. And as the result, it’s very, very good. Like they almost taken the ecosystem from Llama and now they are becoming the default for researchers to do research, which is like a huge branding for Alibaba. And like, I guess like if Alibaba wants to do any kind of business, like it’s quite easy for them.to approach to researchers saying, we are not nobody, right? We are the provider of Qwen and everyone wants to talk with them. And another side for the business is that they find it really hard to attract top talent if they do not do open source, because all these talents want their name on papers, et cetera. if they can pay a lot of money.but they still do not have the best talent. But on the other side, if they do open source and the researchers know that they come to this group and they can have their name marked on history, it’s going to be very attractive. So like this company, even not releasing the best models, they try to release something to make researchers happy. It’s kind of like their...company perk. So that’s another route. But after DeepSeek, everything changed. People know that if I do open source, I can have huge branding for my company. DeepSeek is not doing any kind of commercial stuff, like alteration to cusTiezhen Wangers. Yet they still have a huge evaluation of, I think the most recent number is [unclear: “14 million HKD” in transcript; confirm figure].That’s a lot of money. So by doing open source alone, they can make money. And that changed the mindset for lot of people. so after DeepSeek, Kimi, GLM, Minimax, and StepFun, they all come into this open source world. actually, they have made a lot of success stories, like GLM and Kimi, by doing open source, lot more people understand them. And they kind of open up.the global market, not just the market in China. for them, I feel that it’s not like losing a lot of money because they doing advertisement in a different way. Kimi was spending tens of millions RMB per year on advertisement. And the result is very short retention. People know them, come to their side, and they do not feel any different. And they just move away. Now, the researcher team, the manager, the executive means, knows that the best score on open source benchmark is the best advertisement. So they can concentrate all their power, not wasting them on advertisement, but concentrating all their money and resources on training the best model. But this best self, it’s the best marketing, and they can create great models and start earning money.So I feel that on the business level, everything starts to make sense. But now there is a new challenge, which is how you can stop people from taking the free ride. It’s a longstanding problem for open source. I did something, for example, I made a database. I spent a ton of engineering hours. I open sourced it. But I’m not making any money, because the cloud provider is taking that for free and start making money and monetizing it.it’s happening for open-source world as well. I open-source the model and all these inference providers and chipmakers and BDA-AMD are making money, but not the researcher who created the initial model. That’s why you see some licensing change and discussion on that. Kimi did the first non-commercial license, and then MiniMax made a more restrictive version. Tiezhen Wang (13:40)made a more restrictive version. But I don’t think that’s the final version. People are still trying different things. And I believe maybe in one or two years, we will have a more standard way of balancing open source and commercialization, et cetera. So that’s the second level. The third level is the third level. So the Chinese government is really encouraging people to do open source.If you do open source, you have extra credits on your bachelor education, et cetera. And Shenzhen recently announced a very interesting policy. So you can have housing points if you do open source on GitHub. basically, they are categorizing.Grace Shao (14:21)So the incentive, yeah, go straight to the students, like even in academia, while they’re still in university.Tiezhen Wang (14:27)Yeah, so it’s kind of cultivating this open source culture when other researchers and developers are still in universities, which is really good. So I do feel that the culture of open source is, if they are winning the young students, we are going to see more open source projects. And to be honest, I do feel that that’s the right approach.Because if you’re not thinking about open source, you are thinking like traditional way of collaborating with people, which is company or corporation. And I feel that the essence of why we had cooperation or company is not keeping peace with how we evolve now. I think about, you set up a company in Hong Kong 200 years ago. Why? Because you have a group of people. You want this group of people.That’s why it’s called company. You have a group of people and you want them to work together. And how you can make sure that everyone had their benefits. Everyone is doing a lot of work. Obviously, they want to have a return. And you do that by setting up the shares and also the voting system. that’s how a group of people is working together. But now the word company has changed. It’s more like amulti-international company where the worker in the company has no work in deciding how the company runs. Whereas open source work is more likely the original version of a company. You have GitHub, you know who has contributed what. Everyone knows your contribution, and you can have your name listed. the group of people coming from all around the world, can.collaborate on something. They do not need to be part of a big company going through all the interview process. They can just collaborate. So I think that’s very, very interesting. And now with Zoom, Tencent meetings, and all the Google Docs, it’s much easier to collaborate internationally. I don’t need to know who is contributing to the PR, but I know someone is interested in my project, and we can work together. And I feel that.That’s probably the future way of how people can collaborate. that’s to end the last point on society level. I think the society is advocating for open source. also open source is probably the way how the society will evolve.Grace Shao (16:48)Thank you. is like so insightful pack that I have to digest that. But you you mentioned quite a few different topics, which I can definitely take this straight, conversing different directions to start. have two questions and they’re actually unrelated. So one at a time. Number one is you really make a point about China being really, you know, strong advocate on open sourcing the LLMs. However, I thinkCould you tell us the history of open source in China in general? Was there a tradition to want open source technology even pre-LMDs? That’s number one, first half of that question. Second half of that is you say there’s a lot of incentive for researchers to actually want to open source everything, right? And then therefore they can claim their contribution. Well, in the recent interview between Zhang Xiaojun and...deep minds, Yao Shui Yu, I think maybe you’ve also listened to it. You know, one thing that really stood out to me was how he was saying people need to be like responsible. And like for someone who’s not technical, I actually really struggled to understand what he meant at first until like actually Jiang Xiaoxuan actually asked him to clarify as well. His whole point is that in academia, people are so used to only claiming a certain section of what they contribute. So for example, for a big piece of paper or research,that you would take credit for what you contributed, right? And you want to make sure that it’s best optimized, known, heard, seen, whatever, right? Recognized. However, in terms of how LLM can work properly in terms of the long run, whether it’s like, you know, further in post-training and further, you know, know, usage, whatnot, it’s important that people don’t claim so much credit to their own part of the work. It’s more important that people work collaboratively. But kind of to your point on open source that, you know, they can work collaboratively and make sure that each piece works together better instead of each piece working best on their own. So it kind of contradicts your comment on why people want open source, because in that sense, wouldn’t it make sense for people to not want open source? I don’t know. That’s another question. And the third part of this is really if open source makes so much sense for tech companies and makes so much sense for academics.then why are the American labs so anti open source right now? Like what is driving that? Is it purely because commercial reasons or philosophical reasons? This is very big, but you did throw a lot at me. So I’m going to throw these questions back at you.Tiezhen Wang (19:07)Yes, sorry for my very long answer. I think it’s probably by itself worth writing a blog post with enough content, and I can elaborate more. But great questions for the story. Can you remind me? I guess we can go through them one by one. Can you do mine? Yeah.Grace Shao (19:25)Just like in general, source China, China open source. What’s the sense on that? Beyond LLM, right? Like why did Chinese companies always contribute to open source technology? Clem talked about this in his interview, but he didn’t go into that about it, right? So number two was just about, yeah, number two was just about like, why do these academics want to claim their names, right? Is it better for the company in the end or is it just best for them, like the selfish reasons?Tiezhen Wang (19:37)Yeah, okay. Let’s try it. Yes. Mm-hmm. Yep.Grace Shao (19:52)And number three is why are American labs kind of anti open source right now?Tiezhen Wang (19:56)Yeah, so let’s try to address the first one. I think it’s a great question. And I do see the shift. So I feel that AI is probably one of the very few areas where Chinese open source contributors dominate. If you look back to, for example, I would say the initial days of modern open source comes from like anLinux or Apache or database and everything. And where you do see a lot of individual contributors from China, but you are not seeing enough Chinese company creating a project. And then the project gets adopted globally. You are seeing that gradually when we move to the area of cloud-native, like when the Kubernetes comes out.And a lot of Chinese cloud providers are trying to really pay attention to this whole open source world. And you will see that this grows. But now it’s like this. So it grows exponentially. So I think it comes from two folds. The first one is the Chinese participation in the global market. It needs time to warm up.Like for example, lot of Chinese contributors, they can only contribute two projects in Chinese because of the language barrier. So that kind of limits how much they can actually do. And now with larger language models, with better education in the new generation of developers, the language barrier is not that strong. That’s why.That’s how the Chinese open source contributors can make a better impact. And another one is, so in the traditional way of a company’s, like how a company’s structure itself, if you do open source project, it’s kind of hard to justify your credits because the open source by itself is not the core business of a company. There are very, very few companies whohad their core business made on open source. Like PingCAP could be one of them. PingCAP start with open source and then find monetization plan. But that’s so small. So few of them. And in the new areas, a lot of companies, their core business is open source and plus monetization. Even for IPO companies, for public list of companies, Minimax is basically one such example. They have their best models, open source.and then trying to make money. So this is very different. If your core business is open source, of course you will put more resource on open source. And it’s more likely for your project to gain a lot of developers. And I feel that the third one is the international collaboration has never been easier before, apart from language barrier. So after the pandemic, I feel that all of a sudden everyone is used tolike Zoom and Hangout and collaborating with someone who you don’t see face to face. And this is a great chance for open source project to ramp up. Because before that, you have to meet face to face, and the bandwidth and the people you can meet is kind of limited. And now you have a huge, like, as long as your project is great, like you have a huge pool of potential developers.And the last one is probably AI by itself, like coding agent itself. Although it does make code review much harder because there are probably a of AI scope. But it really lowers barrier of who can contribute to an open source project. Before, you want to contribute to a project. They are probably developing language you do not know. And also, the code base is pretty strong. A developer might not be able to.contribute to the project until he has a very thorough understanding. And that’s probably like months of work. Now you can just ask AI how this part works. And I only need this feature. And what are the code I need to modify? And I can just give it a test locally, and it works. I contributed to some Rust project without being a Rust expert. So that’s how AI makes everything better.So I think it has all these reasons. There are probably more, but I think someone at the end of the day, in two or three years, maybe starting to write some history about how every single aspect of technology, moment and everything, people’s mindset shift, how to cultivate the open source spirit. But I think, yeah, that’s the...Top ones coming out of my mind.Grace Shao (24:26)And then the second question was just that would researchers focused on their own name and frankly ego in this sense actually be the best way to help cultivate the best LLM or whatever whatever product that’s the end product that’s to be shipped. Does that make sense? Because it kind of contradicts what Yao Shunyi was saying on Zhang Xiaoxuan’s podcast, right? He was saying in that sense a lot of researcherswill try so hard to only own what they are working on, but they have less of a sense of responsibility for the bigger project.Tiezhen Wang (25:00)Yeah, I haven’t really read the broadcast entirely. But talking to your point, feel that open source or not, or having researchers name these data on papers or not, it’s actually a game changer. If you are a researcher and you might want to stay in academia, why? Because all the papers you publish is very important to your career. Everyone sees.Like you have published which paper with like who and the paper well like also mentioned that you contributed which part of work are you Look like my corresponding author. Are you the main person or you are just like contributing a part of it? And you there’s h-index to measure the impact of a researcher So like you have all that infrastructure working for you if you stay in academia But if all of a sudden you want to start workingfor company, you lose off that because the company might be able to say, we have a policy where all your open source and all your paper publications, even writing a blog is controlled by the PR team and they have to decide if you can do certain level of things. So your exposure is reduced. You might be very happy because you earn much more, like 10x the salary compared to be an assistant professor.Like after five years, if you ever want to go back to academia, that’s impossible because you lose all of your track record. And with open sourcing, things are very different. The top nailing app wants to hire the best talent from academia. And the people from academia wants to work for the app. But at the end of the day, the two systemIt’s the same system because all the record is public. So it’s very easy for people to come in and come out, come in and come out. It’s kind of different from people coming from academia and then lose track in and get lost in the company world. So I do feel that having your name listed on the work you publish is very important. It is kind of the concept of [Chinese phrase unclear]. So you.your branding grow with whatever you have done. So your reputation is based on whatever you contribute. So you need to pay extra attention on that.Grace Shao (27:16)I see what you mean. And the last bit of just now what we’re talking about was just why is that if you believe open source makes so much sense to these researchers, that so many researchers in the US or at least some of the companies, the entities right now are not willing to go open source.Tiezhen Wang (27:31)Yeah, there are lot of companies who change their position. Google used to be the company which impressed open source the most, like Google open sourced and like TensorFlow Kubernetes and bunch of other important open source projects. The open sourced transformer, which is the cornerstone of our modern AI system. So Google was really impressing open source. But I feel thatAt some point in time, Google stopped doing all the open source work because they kind of lighting OpenAI and other companies taking the free ride. Google did a lot of fundamental work and then it’s kind of taken by other companies for free. And also, OpenAI and Anthropic, I feel that they are still contributing to open source, but that’s not their main project.Their main project are hidden secrets that they do not want to share so that other people can catch up. And the research and business is getting so coupled. So for example, a researcher in OpenAI found a way to improve the intelligence level by 10%. Let’s take o1, for example. They managed to find out how to make model think. And by this chain-of-thought and thinking process, the model islike way much smaller, way much smarter. But they do not want to share the gist.As long as they do, other people will catch them up. So it’s kind of a very restricted environment. Although researchers may still want to open source some of their work and have their name listed there and sharing very detailed observation. But the business doesn’t just allow them to because you are trying to. Yeah. also, researchers can also talk to like in some conference. I know how open source, sorry, I know how o1 was roughly made by reading bunch of YouTube videos from the internet made by OpenAI researchers. But after that, you see less and less very detailed research sharing, even the videos or recordings by them. I guess they kind of learned the lesson.But like, yeah, yeah, that could be part of the rhythm. On the open source world, like, it’s kind of different. Like, so before DeepSecR1 was released, lot of people were speculating how OpenAI was doing o1, and they’re trying things in different way. And after DeepSecR1 is released with all the recipes and all the data they shared, like, the open source world seems to be converged on the path. Although that path mightGrace Shao (29:46)There can’t be compliance reasons.Tiezhen Wang (30:12)be different from o1 because we never know how o1 was made. But then because of DeepSeek’s contribution and sharing, everyone knows how to make thinking chain. And the whole ecosystem is evolving really, really fast. That’s one of the real value of open source because everyone can just collaborate. No one is holding secrets. Well, there are still a lot of secrets on how you can run as efficient as DeepSeek, but that’sLike too technical, like that’s not too much on the research side. yeah, like I...Grace Shao (30:44)So on that note, yeah, I want to follow up on that. I think I recently wrote about something which is, speaking to our researchers, it got me a sense that DeepSeek in a way is now becoming essentially like a foundation for everyone because, you know, a lot of the labs in China are looking to DeepSeek to see if there’s any like, you know, engineering breakthrough, like your point, and they build on top of each other. Help us understand like each of the labs, because you said, they’re cost constraint, they’re compute constraint.Tiezhen Wang (31:06)Thank you.Grace Shao (31:12)They’re teleconstrained, right? Their resources are constrained and every single asset you can think of compared to the American peers. Now, why does it make sense that they all open source and how are they all optimizing for their own goals at thisTiezhen Wang (31:24)Yeah. So open source by itself, as we just talked about, is an accelerator of the whole ecosystem. So DeepSeek shared all their like, knowings and discoveries and what things work, what things doesn’t work. This by itself is accelerating the whole industry, not just Chinese open source, but also like US open source and US like closed source. Like they just don’t say how much they learn from DeepSeek, but I believe everyone is learning from DeepSeek.Not just that, DeepSeek also contributed to GRPO, which has become the most used algorithm, reinforcement learning algorithm in the industry. So they did a lot of contributions. if you check recent model architecture evolution, what’s proposed by DeepSeek is becoming the standard and getting adopted by many people. For example, Kimi 2.5 was using a model architecture very similar to DeepSeqs. And GLM 5.1 was adopting a lot of components from DeepSeek architecture as well. So it’s kind of sharing and learning and co-evolvement is one of the, I would say, secret of how China is able to. catch up with the US in certain area, although having restricted compute and restricted capital, I would say. If US open source is working again, like the whole ecosystem, like everyone was trying to open source, I would say the human race would be evolving much faster than what we are doing now.Grace Shao (33:01)So on that, how do we understand the accusations of what is being distilled? What is technically shared? What is, how do I understand the gray area of that? Like the accusations from a lot of American labs, Chinese labs right now, like you just said, a lot of American labs are learning from Chinese labs. Frankly, within the researcher community, it’s not even Chinese versus US, it’s really just labs with each other and against each other if they have to, right? Intellectually competing. So then how do we understand what the, industry agreement is on a distillation, why is it so contentious rightTiezhen Wang (33:32)On distillation, yeah, that’s a great question. I can only give you my perspective. first, distillation is a very broad word. We are distilling from each other as well. I learn from you, you’re learning from me, and we are all learning from books and papers and all this public information. So I would say,distillation is a very common practice, like basically how you learn from others. Like you might have a model which summarizes the books and like doing bunch of explorations. And the way for the model itself to move forward and evolve is to distill from its historical data and historical experiments. And like that works for like another model trying to learn like your model as well.And on the research field, distillation is very common. DeepSeek R1 was released with MIT license. Specifically, so I actually asked the team about it. They choose MIT license because they want their model to be distilled by others. Because that was the only model that works really well with the thinking chain. And they want all the open source model to be able to have that.Like they have shared all the recipes, but others do not have data. So DeepSeek design like they’re like small models so that and also the recipes so that other people can easily distill DeepSeek, getting the thinking chain and use that on their own models. like this distillation is happening like everywhere. And I think like US companies are distilling from each other as well. Like I’ve seen like the recent discussion on Twitter in public.where Elon Musk and Sam Altman were kind of battle on that. yeah. And if you think about it the other way, so if you do not allow a model to distill, I mean, the output of a model to be able to train a model which is from a competitor, it’s kind of a very interesting point. Like if we say, I’m reading a book.I’m telling you the story. So you, after reading the output from me, which I think of me as a model, you’re reading my summary and you are not allowed to share the summary to others. You have to read the book, the initial book, not using my summary because of the license, et cetera. That’s kind of ridiculous. That’s not how human transfer knowledge in the past a few thousand years.Like I have a very bold argument. I think that anything like generated by AI should not be copyrightable. So like it should be in public domain, like anything generated by AI, because like anything generated by AI is a distillation of like human entire history and everything that human has created. And if you just take that for free and asking other people do not use that.Like it’s kind of a waste and it’s kind of like blocking people from evolving forward. Because like human content do not have this restriction and why you are putting this restriction on something not copyrightable and generated by machine. So that’s something I do not really understand. So I do see there are terms and conditions saying that my model output cannot be used to improve other models. But I don’t think that’s kind of valid.I’m not sure if someone eventually will file something on the court and we can have a case on that. currently, think there are a lot of things to discuss, but it’s not about if we can distill a model or not, but about something bigger. Should the model creator even have this right to restrict others from distilling from their models?Grace Shao (37:18)That’s really interesting. I think that a lot of the discussions in the public space is really about whether you can use copyright work of human output. And then the argument is always like, just you cannot distill because the company said there’s no distillation allowed. But like to your point, there is no actual clear black and white rule of regulation around this right now. And in fact, it’s it’s bit murky. Yeah. Yeah, yeah, that’s interesting.Tiezhen Wang (37:37)I’m not a lawyer, but I can find a clear answer on that.Grace Shao (37:43)Okay, I want to kind of go to China. Like we’ve kind of talked a bit about the big picture. Well, a lot about the big picture. But let’s look at just the China labs. mean, I know that you represent APAC back then with Hugging Face and you worked around APAC, you lived in Australia. But for the sake of this, know, Chinese labs right now probably are the most relevant out of APAC. Do you think I’m missing anything actually on the APAC conversation? Like, do you think anyone else in the region is relevant in this space that we can talk about?Tiezhen Wang (38:08)Korea is doing really, well. Yeah, Korea is really well. Well, the most, one of the best model is probably Upstage. they, initially they create a Korean model leaderboard, like open source version of like model leaderboard. Well, no, no, the leaderboard was not funded by government. The, the, the,Grace Shao (38:10)Yeah, yeah, give us some picture on that. That’s funded by their government, right? That’s their government funded.Tiezhen Wang (38:26)So Korean, it’s actually a very impactful country, but as the other days, there aren’t enough Korean data. Even for ChatGPT, I think until ChatGPT 4, the model doesn’t speak good Korean. So the model was able to speak very good Chinese from day one, like from ChatGPT 3.5, but because of the data volume, et cetera, speaking Korean was always a challenge until ChatGPT 4.At the time, like now, the open source model is able to speak like Korean. So Upstage create a leaderboard. So the way they solve problem is very interesting. They’re not solving problem by solving problem. They’re solving problem by helping others to solve the problem. So instead of creating a model right away, they create.Grace Shao (39:09)I heard about Upstage from VC in Korea as well, but I don’t know the detail about it. Tell us more about who they are, what they’re doing.Tiezhen Wang (39:15)Well, I don’t know too much about who they are, but I only see their open source contribution. I think the founder is a professor in Guangzhou, but he’s Korean and moved to US. Correct me if I’m wrong. I’m sorry. I’m not really up to date with that information. But I just want to call out because I think that’s a very interesting paradigm. For example, if you are a company, you have your own problem you want to solve.Tiezhen Wang (39:42)Like, how do you want to solve it? Like, you are going to hire some people and define a problem and try to use your own people to solve it, right? So that’s the old way. What’s the open source way? Is you publicly define the problem. You have a leaderboard. Like, you might do a private eval or public eval. It all depends on you. the problem is you have to list your problem.publicly and you have to tell everyone that you can contribute to this problem by submitting a model to a URL and we will do evaluation and see how each model is evolving on this area. So they basically have a leaderboard. And you will like a lot of researchers would be very interested because now they have a problem to solve before they do not even know Korean what’s the problem. So now they have a problem to solve and you will see that the curve goes like.it goes up because there are more and more researchers coming in and all their work are open sourced. So a new researcher wants to jump in the field. They will first have a look on the leaderboard to see how far away from a really usable benchmark. And then he can investigate all the previous attempts and find his own way of kind of just changing something a tiny bit.and apply that to the past people’s work and submit to the leaderboard. And now we are seeing people making progress on the leaderboard. So that’s a very, very clever way because it’s not one company solving the problem. It’s like we are opening the door for everyone to come into this playground and try to solve the problem together. I think within a few months, they were able to get thousands of submissions.which is really massive because just imagine you hire 10 % people, you won’t get that. And now it’s by this new way of doing things like building public, evolving public, you’re having a lot more submissions and you are educating people, et cetera. So they have this very impactful and inspiring leaderboard and then they release a model called Upstage for something. I can’t remember it has been a while.And the Korean dataset and the Korean models are accelerating very fast on high-netics. I think it is now the fourth largest models, speaking Korean. Yeah.Grace Shao (42:04)Very interesting. Yeah, I’m going to shamelessly self-plug in. People can listen to the episode I recorded with one of the leading Korean VCs as well that was published last week. He gave a good AI ecosystem breakdown of stuff.Tiezhen Wang (42:12)okay. Yeah, could you help me like do some like DD first and like just make sure that are correct. Yeah, you can.Grace Shao (42:22)Yeah. No, no, no, he did talk about Upstage as well. It’s very interesting. Yeah, I want to... sorry, go on.Tiezhen Wang (42:29)Yeah. And also, so you asked for APAC. So in Singapore, there are a lot of great researchers, like lot of Chinese researchers will go to Singapore as well, like Cancun too. Yeah.Grace Shao (42:43)Yeah, I think the ecosystem is a bit overlooked by I think Western markets, but definitely there’s a lot happening in around Asia. Like APAC has been including Australia as well as Southeast Asia, East Asia, and Northeast Asia. Okay, I want to bring it back to China. We’ve been kind of talking about China kind of more on the high level sense. Now looking at the companies themselves or the labs, we want to break it down. Just give us a sense like, how do we understand moonshot?Mini, Max, Deep Seek, Zhipu, if you have to put it in one bracket, versus the hyperscalers, Tencent, Alibaba, and ByteDance, in terms of their strategy, in terms of the capabilities. Like how should we understand this ecosystem right now? Are there other relevant players that you think I’ve missed, maybe like Xiaomi or anyone else?Tiezhen Wang (43:25)You mean like how the model creator, model lab, are collaborating with hyperscaler? Is that your question?Grace Shao (43:32)No, no, I just think it’s like the people, the people who are creating LLMs, like are researching on how to deploy LLMs. These are the main players, right? Now, how do they defer? How are they similar? What are we seeing like on the ground? Are some of them becoming more irrelevant? Are some of them becoming maybe say, we just talked about DeepSeek becoming almost infrastructure provider for the whole ecosystem.Tiezhen Wang (43:39)Yep.Grace Shao (43:58)You know, are that mini-max is very, focused on multimodality. Zhipu is very focused on coding capabilities. know, Alibaba really trying to push out commercialization by their existing applications. How successful that is, that’s a different question. Just like an overview of these players.Tiezhen Wang (44:14)Yeah, I do think they’re kind of converging. Yeah, because everyone knows that coding is, the whole market for coding is booming. And if you have a good coding model, you can sell it for profit, for large profit. And I do feel that everyone is rushing for coding. There are people exploring different things, like,For example, Tencent is putting a lot of efforts on Hunyuan and doing OCR stuff. And lot of other companies are doing video generation. But at the end of the day, think from a strategy level, I don’t feel that there are a lot of difference. It’s more likely a case where, you have data? For example, it makes a lot of sense for ByteDance and Kuaishou to work on video generation models because they have a ton of data. And also, do you have?like a large enough scale. Like for example, Kimi is not very active in making all the apps. Like Tencent is making models. They are making like great apps. Like for example, Yuanbao, like a QA app, like based on all the Tencent data, it’s very popular. Like they make QClaw. Like Tencent is able to do that because Tencent has a huge talent pool. Like Tencent is a huge company, Whereas like if you look at the Kimi, Kimi is very conservative in...doing all that because Kimi is still a very small company. So I think from a very high level, everyone was on the same page about the strategy. It’s just more, how much resource do you have? What are the advantage of you? Do you have data? Do you have distribution channel? Do you have product design, success story, et cetera? So yeah, I’m not sure if I answer your questions.Grace Shao (45:57)No, no, that’s good. So we kind of talked about why researchers want to open source. We talked about these companies are somewhat doing the same thing. So then this leads me to the question. We know that open source, open weight does not actually mean they don’t make money. However, obviously means that it’s harder to commercialize as we like alluded to with the US labs, why they make those decisions. Then how do these companies find ways to monetize and sustain their businesses then?Tiezhen Wang (46:21)Well, in the US, are also labs dedicated in making open source models and still making money from other donations or from other parts, like selling apps, et cetera. It’s basically the same way in China, too. For example, DeepSeek is run by, I would say, donations from the people who play the stock market.like there are labs run by VCs and lot of labs are already profitable by like selling tokens like GLM has recently raised the token price because like they see a huge number of demand and they’re like running short on compute. yeah, like open source can make money. Like there are a ton of ways for open source model provider to make money. I have a lot of ideas. if in case you are interested in like making yournot profitable, can contact me. But honestly, there are lot of ways. The simplest way is to sell token. If you have the best model, you can sell a token for profit and people will actually buy your token. so it’s very interesting because when we combine science and technology, always consider it’s the same thing.Grace Shao (47:15)Yes, everybody find Tiezhen Wang.Tiezhen Wang (47:37)For model, it’s the same. When we think about models, we just think of a model that generates tokens, et cetera. But actually, there are two different parts. The first one is training, where you have the model. And after you get training, open source the weight you trained. Another part is the inference. So you need to run a lot of optimized CUDA kernels in order to make your token cheap and fast.Either bracket can make a lot of money. For example, you can open source the fine-tuned model, not the base model. So if a company want to use open source model for fine-tuning on their own data, they cannot be building on a fine-tuned model. cannot build. They have to find the base model. And if the base model is not open sourced, you can sell that for profit.And also different clients might have different requirements on the model. The NeoLab can collaborate with the client directly and provide some kind of training and post-training support. So that’s a way of making a lot of money, actually, because training is very expensive. It involves very expensive researchers and data and compute. On the inference side, too.Grace Shao (48:45)Yeah.Tiezhen Wang (48:52)Because the inference is tightly coupled with the data center you own. So your optimization strategy does not, there’s no guarantee that your optimization will work on a different cluster. So a lot of people just do not open source the inference recipe because it’s not that useful. And also it’s kind of a moat. So the model provider who creates the model, they know how to optimize the model best.when the model is released because they have seen the model for four months and they have done a lot of optimization on the model inference. And when the model is out, like everyone else, it’s just starting to know the model and doing some optimizations. So of course, the model provider will sell token in a much efficient way compared to all other competitors. Three months later, when the outside inference provider gets toknow all the secrets and do very optimized kernels, there’s a new model coming up. So the model maker, the people who know the model from day zero, always have an advantage on selling the tokens. So that’s one of the very important ways how they can make money.Grace Shao (49:59)I see what you mean. Mm-hmm. Yeah. And does DeepSeek v4 coming out have an impact on how the GLMs of the world or Kimi make money? Like essentially their strategy with the fact that you just said they just raise prices on their tokens.Tiezhen Wang (50:16)Yeah, so GLM and Kimi doesn’t sell DeepSeek or Qwen. So they are not competing with each other directly. I would say the capabilities are on par with each other. So it’s more like a user test. Which one is better? There is no clearly winning between all three models. So we’ll see like Zhipu’s stock price was getting down because people were so worried about DeepSeek. But then they realized that like theZhipu token selling is not quite impacted, so the stock price bounced back. But at the end of the day, I would say it’s actually a good thing for them. So GLM 5.1 is adopting a lot of core design in DeepSeek with 3.2, I think, model architecture. And they were able to cut down the cost.by adopting all these exploration from DeepSeek. And now V4 Pro is out. I don’t know the details, but a very simple guess is that Zhipu is able to cut down the cost because they can adopt new things from DeepSeek architecture. So Zhipu on one side, because of the demand is so high, so they can increase the token price.and they can learn from DeepSeek and cut down the cost. So Zhipu is going, yeah, exactly, exactly.Grace Shao (51:34)you have a higher immersion. Yeah, this is something I think they’ve talked about as well, like really being able to learn from the engineering breakthroughs that DeepSeek puts out every time. Okay, I have mindful time. I just kind of want to have a few questions on the future outlook. You posted on X recently saying that you’ve been thinking a lot about how do we make AI bootstrap itself? And you you’re going through this transition yourself, you’re thinking about the future of AI. What does it mean for the open source future as well?Tell us a bit about where you stand right now and how you think of this bigger picture.Tiezhen Wang (52:06)Yeah, I’m still doing some exploration on my side. I think this whole AI bootstrapping logic has already been implemented by a lot of big lab internally. The idea is very simple. In compiler world, you can design a programming language and write a compiler probably in well-known languages like C. And then you will first implement this language using the C code.In the next iteration or after a few iterations, you are able to implement this language using your own language. So it’s called bootstrapping. You are basically evolving on your own. You are not relying on something which is not from your language. So it’s like putting it another way. If you see how normal living creature, how they replicate itself and how they evolve.I don’t need to have a screwdriver somewhere to engineer my kid, right? My kid’s just born. All by itself. But how far are we from AI to do similar things? Now we have a coding agent very powerful. We have our AI training pipeline recipe kind of stabilized, at least for small sized models. So are we really far fromlike AI able to get one of my idea, like I give him the direction, and he’s able to like first bootstrap a very simple version and gradually evolve towards that goal. Like I think it’s like highly possible. So at the end of the day, we might be able to like just tell him what I’m going to do without like giving him all the harness and all the like detailed guidance and.I’m not talking to him 100 times, and he’s able to first lay out what he needs to do and have a plan, and then probably design a DSL or agent all by himself. And probably he will create a model ways to help him to get adapted to this goal. And then he can just keep evolving. All I need to do is to give him more fuel.which is compute, and he’s able to do some evolution and all by himself. It’s kind of like if you have recently read Andrej Karpathy’s Twitter, there’s a concept called auto-research. But auto-research is just evolving on the model weight. It’s not evolving on the agent and harness. I think on the agent level and harness level, there are also a lot of things to do too.So I’m quite new on this journey. What I was able to do is to bootstrap a very simple agent and I can use that agent to optimize the agent. But I think eventually we will get the weights involved too. When the model realized, okay, I’m not just needing an agent, I can create a bunch of data and improve my weights. He’s able to evolve from that too.Grace Shao (55:05)So in the future, how important is the capability of the models versus the harness and then the industry expertise then? Because right now, so much of conversation is still about, you know, the models are very strong. We are seeing what you’re saying already, the agent’s starting to build out things auTiezhen Wangatically. But we still need the taste. We still need the industry expertise to guide them. I find it hard to imagine that, you know, you can plug in something just say, want this to be done. And the agent just starts doing it exactly to your taste and your...imagination? Do you really think that’s happening?Tiezhen Wang (55:34)Yeah, I do feel that it’s happening. Like we are using agent to like especially coding agent to code something that we are completely unfamiliar with. And I’m quite confident that it will actually work. The reason is that like I have defined a set of goals and as long as I see that is moving towards that direction, like I’m good. I do not need to understand the code line by line. Like it’s just the box. But like the difference is I’m using the coding agent.Grace Shao (55:58)Mm-hmm.Tiezhen Wang (56:02)to do something else. What I can do is to use the coding agent to improve coding agent itself. And using the coding agent to generate the data and train the model that coding agent is using. And I would call that a bootstrap, not like just using, like, I think I’m already quite happy with coding agent to do something else. But just like, yeah, yeah.Grace Shao (56:23)Interesting. I want to end on a more philosophical note. So do you view the argument that AI is going to replace humans then? Or do you think AI is going to be in the role to support humans if we could keep on going down this path?Tiezhen Wang (56:35)Well, it’s actually a very, interesting question. And I feel that people do have different feelings. But from a pure technology point of view, I do feel that it’s one condition of like, so technology is not the only thing that will decide everything. You mentioned that if it’s going to help humans, well, it’s not really technology by itself to decide.Like it can be used in different ways, in different social structure, in different like tradition and such. It’s like giving you a gun and you can do things in good. Yeah, it’s not. Yeah, yeah. But like just imagine that you’re going back to history with all your knowledge of modern society. Are you going to help theGrace Shao (57:12)It’s not a good analogy. But yeah.Tiezhen Wang (57:26)like the society, like the history you go back to? Or are you able to help? Like I think it’s basically the same. If you have AI that knows everything, you can just think of it as human being in like 2000 years in the future. And now you have it. And what it is going to help on the society. Like it really...Grace Shao (57:31)Yeah. It’s like that saying, your own capability of using it is the cap of itself. Also, I think there’s a lot of argument and discussion around the fact that even the society as we know it today, the knowledge work that we all have, that we normalize, are not even created until the recent 100 years. And if AI is to disrupt that and replace human in that sense.Why is it so bad? Because it alleviates us to do other things that human multifaceted beings that we are can do. Is that kind of part of the argument as well where like, even if it replace us or helps us, it’s only helping us actually alleviate some of the things, if you take a step back, the things that we don’t want to do, right? Where we can maybe go touch grass. I don’t know, maybe this is very optimistic view of it, but there has been people saying like,The cap on AI capability is a cap of your own intellectual, your own cap of your own ability to navigate or use AI. So the more you can use AI, the more it can help you. The less you can use it actually, the more it will replace you.Tiezhen Wang (58:45)Well, I think it’s very interesting to define what is you. Are you defining you as everyone, or are you defining you as people who have compute? Well, no, it’s not. It’s actually a very, very, very interesting question happening right now. You know, Anthropic coding agent is able to do lot of things. But people are of imagining that we areGrace Shao (58:53)We’re getting really philosophical now.Tiezhen Wang (59:09)Everyone is getting a lot more powerful with models. But what if one day Anthropic just say, you cannot use your coding agent to do certain things? It already happened. Anthropic said, you cannot use your agent to do auTiezhen Wangated tasks. The other thing, there could be other limitations. I have a very bold argument is that the reason why we are able to use AIso cheap that even us, like we do not own a data center, right? Even us can use that. It’s because our data is still valuable. You know, if you use subscriptions, your data is going to be distilled by Anthropic to further improve the model. And like they are able to give us a discount because they still need our data.Grace Shao (59:52)So your point is that one day when they capture enough data, they will not even give us this kind of access for free or for cheap price.Tiezhen Wang (59:59)It depends on how they define you. You ask them, like, you or something. How they define their user. How they define who could be part of the game. Like, if one day, like...Grace Shao (1:00:09)So then the question is, no, but then my question is, then there’s another argument where they’re saying too much power is in the hands of a few companies right now, right? Or a few founders, what not. We need open source, that’s your point, right? No, that’s really interesting. And that was actually gonna be the last question I was gonna ask you. What is one differentiated we hold? And I think you’ve already answered that in that sense, right? Yeah, I think it’s for us to really think about it. But then as the average user, my question is, how do you actually boycott?Tiezhen Wang (1:00:18)That’s why we need open source.Grace Shao (1:00:36)these companies or if not boycotting, how do you actually make an impact? Because if I’m not the developer creating an open source model for the average person to use, me as an average user, what do I do?Tiezhen Wang (1:00:47)Well, just use the model to do the thing you want to do. Try to embrace the model and be more patient for open source models because obviously the open source model is not as good as top tier closed source models. you kind of like, well, I mean, with open source models, you keep all your secret to yourself. So you can have.like better security and you have better control. Open source model will never betray you if you just write on your local laptop. So although the model is not performing as well because he’s not distilling you, right? So still you can trust on your open source models and give it a more task to do.Grace Shao (1:01:20)You host yourself.Tiezhen Wang (1:01:34)I do feel that a lot of open source model is actually capable of doing things. But the expectation might be, think of it as six months, like cloud version of, sorry. Let me put it another way. So think of it as old closed source models and be patient with that. And you can grow up with the open source model together.Grace Shao (1:01:54)That’s very interesting. Thank you so much for your time, Tiezhen Wang.Tiezhen Wang (1:01:56)And thank you, Grace.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Nathan Lambert Reflects on China’s AI Labs: DeepSeek, Open Models, and the 'Race' with the U.S. 19.05.2026 1ώ 3λ

  Joining me today is Nathan Lambert, author of Interconnects AI and a post-training lead at the Allen Institute for AI. Nathan recently returned from a major tour of China’s leading AI labs, where he met with researchers and teams building some of the most impressive open models in the world.In this conversation, we discuss what Nathan saw on the ground: how Chinese AI labs differ from their U.S. counterparts, why open models have become such an important part of China’s AI strategy, and how labs like DeepSeek, Alibaba, ByteDance, Kimi, Z.ai, MiniMax, and others are navigating compute constraints, data access, and commercialization.We also dig into some of the most debated questions in AI today: Are Chinese labs really 6-9 months behind U.S. frontier labs? How meaningful are distillation accusations? Can domestic chips like Huawei’s make up for restricted access to Nvidia GPUs? And is China’s AI ecosystem actually government-directed, or is the reality more fragmented and commercially driven?Ultimately, this episode is a more nuanced look at China’s AI ecosystem that looks beyond simplistic narratives about subsidies, copying, or geopolitics, and instead examines the technical, cultural, and economic forces shaping the future of open models.Check out his two recent articles here:* Notes from inside China’s AI labs* How open model ecosystems compoundTo find the previous episodes of Differentiated Understanding, see here.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently. Season two will host a series of guests from early-stage investing, as well as builders, researchers, founders, and product managers. For more information on the podcast series, see here.Chapters00:00 Insights from the China Trip11:51 Cultural Differences in AI Research18:15 The Role of DeepSeek in China’s AI Ecosystem25:26 Overview of Major Chinese AI Labs30:56 The Future of Open Source in AI37:50 Market Dynamics and Consolidation in AI42:28 Distillation and Model Convergence Controversies51:58 The Gap in AI Performance: US vs China61:09 Monetization Strategies in AI: A Comparative Analysis62:32 Government Influence and Misconceptions in AITranscript (AI-generated for reference only)Grace Shao (00:00)Nathan, thank you so much for joining us today. Yeah, really, really excited to finally hear your thoughts on your big China trip, on what’s happening between the Chinese AI labs and the U.S. AI labs, what you think the potential compute constraints might mean for these labs and their performance in the future, and obviously the open-source ecosystem. So before we get into all of that, could you...Nathan Lambert (00:02)Yeah, thanks for having me.Grace Shao (00:23)Briefly tell us about how you ended up actually working on post-training and open language models. Just a bit about yourself.Nathan Lambert (00:29)Yeah. So I actually started my PhD at Berkeley in 2017, not working on AI things. I was an electrical engineer by training in undergrad, which is funny looking back, because that’s the same year that the Transformer paper came out. And I was like, I think I should do this AI thing, and tried to get the famous advisors to mentor me. And they’re like, we can’t take you. So I had my PhD as this wandering path to become an AI researcher. And then I ended up at Hugging Face after that, which was, realistically, the only industry research job that I had, but also a very hot startup and very fun to learn kind of at the intersection of these tools that people use a lot for AI and research, which is what I was doing.And then when ChatGPT hit, the kind of RLHF thing blew up as the hot word on the technical side of things. My PhD had ended up being in reinforcement learning, which is just the first half of reinforcement learning from human feedback. So it was kind of a natural pivot to be like, well, I might just do that. And Hugging Face was a good place for doing that, because the whole company is kind of all for that, which is like: figure out how to support the community on the hot thing and build platforms there. So they were very happy about that. And I helped build a team at Hugging Face.And then I was kind of burnt out on the remote-work time-zone thing and found out that the Allen Institute was doing such similar stuff. And I was like, wow, I have people that could be in-person friends and do similar things. I was like, quality of life — I need to do this. And a few years later, I ended up building a bunch of models. And I think being at a nonprofit opened me to this ecosystem vacuum of information, where there aren’t many people who can talk about what they’re doing. So then, with some luck and committing to write every week, I just feel like my influence filled the vacuum of nobody saying reasonable things.And it is this nice synergy between what I write about and what I work on in my day job, and it just kind of got bigger and bigger in a very fun way. I think that, generally, at the highest level, I’m motivated by wanting AI to go well on this trajectory. And I worry about a lot of near-term things, whether it’s social unrest in the U.S. and just kind of the massive hatred for AI — I think is a very big near-term problem — and then, medium term, concentration of power, because I think AI will be super powerful in ways that people don’t expect. So generally, open models are a nice way to curb both of them by being a bit more transparent to people, and it naturally is a hedge against concentration of power. There have been different reasons throughout that, but that’s kind of a recurring theme in my life in the last few years.Grace Shao (02:50)Definitely. I love your work because I think you help non-technical people like myself really understand what’s behind what’s happening in these labs a lot better. And then I actually just spoke to your former colleague, Tiejin Wang, and he was with APAC Hugging Face just last week. He was saying the same thing. Open source, in many ways, is kind of the best way to go forward as we know that this technology will not stop evolving, but it’s the best way to kind of put up guardrails and checks and balances for the monopolies.Okay, I don’t want to take up too much time on that side of things today because our focus really is about your China trip. Before we get into the weeds of all that, I want to hear about the trip itself. Most people who are writing about Chinese AI are getting their information secondhand. You really went there, you spent time with the researchers, you met with people who are building the models. Tell us about what you meant when you said you came back with great humility, right? Your eyes are a bit more open, whether it’s the good or the bad. Tell us about your trip.Nathan Lambert (03:50)I feel like I kind of went in — I mean, I had this horrible English phrase in my writing, which was like, “I knew I knew nothing about China,” which kind of tried to indicate that I knew going into the trip that I knew nothing. And it was still the fact in my current writing. This is a horribly written sentence that I had in there. And I only talk about it because somebody called me out on it. It’s like, what is this? And it’s like, leaving, which is knowing that it’s such a big country, there are just such vast amounts of talent working on these problems, and how unpredictable it is as a human to model people with very different worldviews and upbringings and training systems. Realistically, the way that people are trained in China is very different.And I just think that even being there, you can’t fully grasp: what are the pockets of three to six researchers doing that is actually a bit different than in the West, even if they’re working on the same goal? I think you could get down to that level of granularity and a sociological study and actually see differences in what they’re working on, and that’ll always change the output. I didn’t get to that level of granularity, but it’s just to start having real experiences and understanding how people explain how they work on these problems.And for me, realistically, a lot of it is coalition building, which is just like: I want there to not be vitriol at the level of the technical companies doing things in international bodies. So just meeting all the labs on both sides is really nice, because you need to do that for them to talk to you about more sensitive issues in the future. I got some criticism on the piece, which is like, this is how you shouldn’t visit China. And it’s like, well, what are you going to do if you’re going on an official visit to a bunch of companies? How do you expect to get in the door without being nice? You have to start somewhere, and I think it’s important to be respectful.Grace Shao (05:31)I think the piece was, frankly — I don’t think the criticism was fair, to be honest, because I think you were really transparent with the fact that you’re not a China person, right? It’s not like you’re going there and exoticizing everything. And if anything, a lot of people, even with China backgrounds, like to use certain dragons and tigers to describe things. I feel like you actually were really humble going and being like, I’m just a technical dude meeting with these labs, talking about their technical research, right? And then because you were physically there, you had observations of the culture and the people. So yeah, I actually thought your piece was quite good. And yeah, sorry.Nathan Lambert (06:05)I agree. I was willing to let that sail past, but I think it’s important for people who listen to realize how actively these companies are trying to court Western audiences, which is why we could get in the door. I mean, we had some prominent people on this trip, but that’s why we got all of them in the days that we wanted them, except for DeepSeek. So essentially some, like Catherine Rintel, who works with me at Interconnects, and some other creators...Grace Shao (06:23)How did you get everyone? Yeah, how did you get everyone?Nathan Lambert (06:29)He used to live in China and has connections in China. So he kind of orchestrated the mix of his connections and leveraging my connections to labs. We had some bigger names on the trip as well. Just stringing all of these together to get all the various labs in place is a few months of networking to make sure the trip lines up with people with established networks and contacts with the various labs. But these people want to look good to Western audiences, so they’re only going to say yes to the right researchers.And the researchers know that there are two to four comms/ops people in the room, hanging out, making sure that it goes well. Especially the bigger the company, the more comms people. You go to Alibaba and there are three to five various people, from the head of comms to some special offices. You’re not going to get these people in the office, or at all, without accepting the cost of these types of handlers. It’s the same thing in the U.S. You’re not going to just plop a senior executive into a chair.So it’s also good because now I have the WeChats of a bunch of researchers from China that I could just text about things. It’s like, hey, congrats on the new model release. It’s like Lay Lee works at Xiaomi, Xiaomi MiMo. It’s like, talk to this guy for an hour at a mall — I don’t remember the name of the tea store — but it’s like...Grace Shao (07:27)No, of course. No, of course.Nathan Lambert (07:49)Now we have these relationships, which is very useful, and that helps information spread across the ecosystem to these trusted parties, which doesn’t really exist. There are not that many, I think. And the opposite direction of the trip is very hard because Chinese researchers can’t really enter the U.S.; the visa purgatory is too complicated. A lot of us on the trip were either Canadian or entered on a transit-without-visa entry, which makes it very easy for American technical talent to go to China right now, which is why I think there are so many trips. I think there’ll be more of them.We’ve got a lot of inbound from VCs and open-source labs in the U.S. that want to establish collaborations with these various labs because they’re the best open-weight models, and they want to build a stack for companies in the U.S. building open-weight models. So I think there are going to be more prominent, but not gigantic, U.S. startups going to try to build these relationships, which I think is a really interesting technological development because we’ve never seen this type of professional work trip in China from U.S. tech companies. Most tech companies have a “bring a device to China, it auto-bricks itself, and you have to hand it into IT.” So to actually proactively send people in a professional capacity is a really big change. There are a lot of angles you could take this, and I think it’s cool to see how it unfolds. This isn’t even really about the trip. This is the follow-on that we’re hearing from people that are like, hey, how’d you do this? We want to do this trip.Grace Shao (09:06)Yeah, definitely. Actually, from my end, I hear about VCs or investors always being quite active going to China because previously American funds were very, very active during the internet era. People were kind of always trying to find a way to either get into these good deals or potentially keep their pulse on it. But I think it’s really, really positive for the whole AI ecosystem to have this kind of fair, transparent exchange in some capacity. But to your point, there’s no way that star researchers can come out and talk to you off the record without any compliance, because that doesn’t happen in the U.S. either. That’s just companies protecting themselves.I just think your trip was quite meaningful, and I want to bring it back to your observations. You talked a lot about the cultural aspects of it. You talked about how you felt like in China there was less of this star-researcher celebrity status around people. People were more humble, or there was more humility. It was very focused on execution. You argue that Chinese labs are particularly well suited to the current LM-building game because they’re very focused on meticulous stack-level work. And there’s less ego sometimes to work on the dirty work, or the non-sexy work. So kind of unpack that for us. Why do you think that is? You kind of touched on it — you said they were brought up differently, they were taught differently — but what’s so different?Nathan Lambert (10:27)So essentially, an interesting part that synergizes on this trip is that we stopped by some academic institutions. I think it was like AIR and Tsinghua and stuff. And you hear all of these academic leaders talk about how they’re pushing hard to try to change it. So yes, they know China is producing more papers than anyone else, but they still think that it’s not as transformative of research. And they think that they’re trying to cultivate the academic domestic ecosystem to change just the type of work it works on, and the distribution, and take more risk.And then you would talk to some industry leaders off the record behind closed doors, and you would hear things like, it’s never going to change because the education system is so structured. There are so many layers of the funnel that reward things like memorization and stuff that they’re just like, this research culture is not going to emerge. And then the follow-on with the AI labs is that these labs are doing fast-following. They kind of have a proof of concept, and they know what it needs to look like. Therefore, in that domain, you’re not trying to invent the new paradigm. You’re not trying to make the model that is o1 or o3, or the first model to work in Claude Code. You’re like, I see it, and I’m going to try to do that and make it the best thing. And I’m going to try to make it cheaper and just maximize that goal.A lot of companies don’t need to invent the new paradigm. OpenAI has done this so many times. That’s their bread and butter: never doubt OpenAI’s ability to release a blog post and a plot that changes how people think about AI. I still think it’s going to happen a few times in this massive boom over the next four years. OpenAI just kind of has that sense of what is the thing that you can push on a bit earlier and just transform things. But I don’t expect — and other people wouldn’t expect — the Chinese companies to do that as much, because it’s just such a culture of, I guess, building. I don’t know how to describe the positive version of this. Maybe it’s slightly more practical-minded, in terms of: it’s your job to build this thing.A lot of the researchers, maybe because they knew their managers — some of them had managers in the room — see their role in the company as being to make the models excellent. And especially for students, I work with students and that’s what they say. I work at the Allen Institute and we have students that will co-lead our language models. It’s not that surprising, because if you do an industry research job in the U.S., a lot of mentors will tell you that you’re kind of free of the burden of bureaucracy and politics. So the naivety of students, and the simplifying, is actually so good at just getting a lot of technical work done.There’s also the life-stage side. If you’re younger, you don’t have as much family, and you normally haven’t built up as many habits and other things you do with your life. Language models are so complex, and the amount of context that you need to absorb to understand what the bottleneck is — there’s so much information, and you have to be able to pick what the bottleneck is and break it. If you just don’t have the mental space to absorb all the context, you kind of end up doing things that are cute but don’t make breakthroughs on the model.So that’s kind of a difference that I’ve seen in people who were both very successful academically before language models. Some of them are able to pivot to this practical mind, which is: what is the state of the system? How do I improve it? And then some try to make kind of these abstract frames of what’s happening and approach it like an academic, and it normally doesn’t improve the model as much. So I just kind of see, if the academic system is a bit more practical-minded, a bit more structured, and the work you’re doing is structured in the language model — make this kernel implementation faster, make this idea work — then maybe it can be...I think it’s an oversimplification. I push on that a bit in the piece just to really contrast what you could think a U.S. lab would look like. And I have a few anecdotes. I’ve heard a U.S. lab paying off a researcher to be quiet about their thing not being in the model. All of these one-off things are more storytelling devices than anything, because most one-off things don’t matter at all. But also Llama 4 imploded, and that was because it was described as a Game-of-Thrones political-style environment, with all the VPs vying for influence and showing that their thing made the benchmarks go up. It kind of fell. Many, many people will tell you that. And we’ve had the Qwen turnover, but it doesn’t seem like it was quite the same type of thing as Llama 4 or xAI. xAI barely exists now. There have been some dramatic things in the U.S. with how these companies have kind of come and gone out of the fold.Grace Shao (14:55)Yeah, I kind of agree with you, but also I would push back on that. I think there’s obviously a more rigid and competitive academic system, which by default in East Asia results in a culture of students following the bureaucracy and authority a bit more. So I agree with you in the sense that they’re very pragmatic. They focus on the task that is given to them. However, I wonder if things will change with how AI will disrupt education. That’s number one. But also, a lot of the young researchers that you’re working with today seem quite different. At least a lot of the entrepreneurs I meet today are born in the ‘80s and ‘90s, some even younger and born in the 2000s. And I think there’s a kind of aura or confidence coming from them. If anything, you want to say they’re a bit more individualistic-minded. You went to Shanghai, right? They are dressed very, very uniquely. They have these outrageous outfits on the streets. People are seeking individual ways to showcase their personality. So I wonder if that will shift.But for sure, for the academic institutions like the Tsinghua and the Beida of the world, they are still very old-school. But I would say that is the same maybe in some academic institutions in the West still. Okay, I think on this topic we can go off on a tangent on academics, but let’s go back to China’s ecosystem.When DeepSeek V4 came out, we talked about it offline, the two of us, quickly about a piece I wrote saying how DeepSeek is starting to look a bit more like a base layer for China. And if anything, some of the labs kind of admitted to that. They’re like, we have very limited resources. And to your point earlier...Nathan Lambert (16:11)Yeah, you could take that in so many tangents.Grace Shao (16:34)Limited people — these labs are tiny. They’re run by 100 to 200 people max. Limited capital, obviously limited compute. They have constraints all around. And in that sense, in a way, the ecosystem’s looking less like a zero-sum game and more like different players optimizing their own strengths. So correct me if I’m wrong, but DeepSeek is providing a base layer where a lot of labs will quickly follow and basically adopt a lot of their engineering breakthroughs. And then Zhipu, Z.ai, will focus on the coding; MiniMax focusing on the multimodality, et cetera. There are a lot of these different players. ByteDance, obviously, very, very focused on their video models. And Qwen, like you mentioned, had the whole open-source saga break apart with Lin Junyang leaving. But in general, they’re still kind of the leader in hyperscalers on that front. So everyone’s doing their own thing almost, instead of really...Nathan Lambert (17:27)I agree with the people specializing, which I think is normal business evolution. You figure out a bit where you’re good at. And there’s so much opportunity that they are like, okay, I’ll follow this because they see that they’re good at it. I just am more skeptical of DeepSeek as a base because I have no idea what DeepSeek is doing. And some of the labs when we were there, because DeepSeek V4 had just come out, were like, yeah, we look at the things they’re doing, but they seem more intricate than needed. And if you read the paper, there’s just so much going on in this model. As a researcher, I’m like, some of it seems a little fake or a little dependent on their setup and not necessarily going to work in every model.Grace Shao (18:04)What does that mean? Break it down for me.Nathan Lambert (18:18)Essentially, I will say that building an LLM is dependent on where you have your GPUs, your pre-training dataset, your intended deployment setup, and stuff like this. So you make decisions based on your constraints, and you build the model. DeepSeek has these constraints and they end up with their model, but Moonshot and Zhipu have different constraints, maybe more flexibility, and they ended up building a different model. They will test the DeepSeek innovations. So they’ll say things like, X innovation doesn’t improve our model. These two organizations are on different development paths that have core similarities, like these large mixture-of-experts models and the general methods are similar, but a lot of the parts end up being a bit different.That’s why I’m like, I don’t know exactly. If DeepSeek was a base, you would see the Chinese labs just do post-training. We just take the base model that’s out there and we adapt it to our domain of specialty. And we have users that do that, which is something that I think about a lot. I’m thinking about starting a post-training lab and how to format post-training research better. So I think about this a lot. I think about what a shared base actually would be. They go through — some of these labs put an extreme cost on creating their base model. And if they didn’t need to do that, they wouldn’t.One of the labs told us how long their pre-training run was, and my jaw dropped. I was like, that’s way too long. Any U.S. advisor would be like, you’re taking way too much risk on this pre-training run. If they didn’t land that pre-training run from one of these past big MoEs at a Chinese lab, I don’t know if the company’s dead, but that’s a huge amount of time. Most U.S. companies now know that you don’t want your big pre-training run to be more than a few months because it’s just so much risk and time to put all your eggs in that basket.That’s a sign that, in that case, they don’t have as big of a peak-size cluster. Essentially, pre-training time can come down a lot when you have a bigger overall cluster; you can just get more throughput on it. But if your biggest cluster is smaller, it’s harder to get a certain amount of throughput, so you use that one for longer. That’s a compute constraint. To loop it back, I think the specialization is real, but I’m more like, I have no idea what DeepSeek is doing. I know they’re raising money now. I don’t know what the plan is there. They seem the most without a specialty in the Chinese ecosystem.Grace Shao (19:59)Dependency on. Yeah.Mm-hmm.No one knows, though. No one knows. They’re secretive.But that’s my point, right? I feel like they’ve been kind of nationalized, whether willingly or not, because they’re taking the Chinese government’s money. They’ve kind of gone secretive. And it’s not like there’s a secret that they prefer Chinese-educated researchers. They’re keeping a very domestic stack, from talent to capital to the whole stack. So to me, it seems like they’re being Huawei’d, in some ways, because they did well and they got their name globally, and then by default they’re becoming the next Huawei, willingly or not.Nathan Lambert (21:01)I don’t think nationalization makes you a base for the other companies, at least not at this stage. There could be something, but it’s hard to force.Grace Shao (21:06)But then you have some incentive, right? But then it is some incentive. You’re like, well, if you can propel one of the teams and propel the whole industry as a whole, it could be in your KPI or some kind of unspoken expectation.Nathan Lambert (21:17)The coordination problem is so hard. Essentially, both in the U.S. and China, even the open labs, what they do is they fork open-source code and match it to their internals, and every company does this. Therefore, all the improvements that could potentially be going to the open code and forming this base that is far more efficient — they’re not completing the feedback loop. I think China could be closer to it. If people really lean into DeepSeek as a standard architecture and DeepSeek shared their training code and all the specifics and how to do this, from a Chinese economic perspective, that would be a huge win because you’re just saving compute. But I think it’s too decentralized and too competitive to have that happen. It wouldn’t happen in the U.S. either.Grace Shao (22:04)It’s so cutthroat. Yeah.Nathan Lambert (22:08)Even though I think for open models to be closer to the frontier, it would be better. I talk about open models in the U.S. needing a consortium. But there’s definitely enough money to make a consortium in the U.S.; then you fail because the model won’t be good because you’re feeding too many asks into the model. That’s the only way to create a shared base.Grace Shao (22:25)Interesting. So it’s not really just commercial. Yeah. It’s not the commercial reason.Okay. So if you had to give a high-level commentary on each of the major labs, what would it be? If you look at ByteDance, Alibaba, Tencent Hunyuan, if they’re relevant, DeepSeek, Moonshot, Zhipu, MiniMax, Meituan, Xiaomi now being part of the ecosystem too.Nathan Lambert (22:46)You might have to prompt it or say more, but I could just kind of ramble through them, which is kind of fun. Alibaba: cloud-focused, understands that open models can enable more usage of platform. So I would say Alibaba is very, very cloud-focused. ByteDance: mostly characterized by everybody else being intimidated by them, and very user-focused, including multimodal. Kimi: vibes of the office were great. It would be one of the best startup vibes that you would visit among U.S. or China. Zhipu: very AGI-pilled, surprisingly cautiously excited about being entity-listed, even though they have no idea why they are, because they’re like, it stamps them as a big deal. And then there’s some...Grace Shao (23:27)I think they previously worked with SOEs. That’s the main reason. Or they still do, but that was one of their main sources of income. And unfortunately, because a lot of these labs spun out of Tsinghua, and Tsinghua is, for people’s context, in Beijing. It’s really close to the government, obviously. But the thing is, when it’s close to the government, it could mean there are three layers of agency underneath the actual government apparatus. But then people like to link it to the fact that it’s taking government money, so therefore they are suspicious. It’s very unfortunate, I think. A lot of companies get thrown into that category. Even companies like Lenovo and a few other Chinese companies have previously been called out by U.S. senators saying, they’re taking Chinese government money, but really it’s that their scientists or their research labs spun out of a certain government-affiliated or government-funded academic institution. That’s what it is. Anyway, yes, go on.Nathan Lambert (24:23)Yeah. Some more would be: Xiaomi — surprisingly great research vibes for a new team at a random company. They seem to be crushing it.Grace Shao (24:31)What do you think of Luo Fuli? The star researcher.Nathan Lambert (24:31)I didn’t get to meet her. I think she’s as close as they have to a star researcher right now. There’s the tier of star CEO, which there are obviously others — Dario and Sam, the analogies are there — but the star researchers, like the Sholtos of the world in the U.S., obviously you can come up with many more. She’s the closest you have to this. I need to watch more interviews. We’ll see. But she wasn’t in our meeting.But they just seem to be doing the right thing. They’re making general models. They don’t really have specialization yet. Florian, the person who helps me write about open models on Interconnects, and I took a detour to go see Meituan because we’re like, why is Meituan building these models? And they’re very practical about it. It was a less glamorous visit at a normal tech office. It wasn’t an official visit for them. They were like, yeah, we’re a major online platform. We obviously are going to use LLMs everywhere once we need to build our own LLM and specialize it to our products, which, surprise, is very practical-minded. I’m guessing there are many more companies in China like this.Grace Shao (25:39)That’s what Tencent’s saying too. It’s because they want to serve their existing consumers and optimize their LLMs for their own distribution and their own basic interface or activity loop.Nathan Lambert (25:52)Yeah. After I left, some people in the group went to Xiaohongshu, like RedNote, and they’re there. They’ve released some language models that are multimodal. They’re like multimodal data-processing things. So a lot of them are not that surprising. The startups just have different cultures. I have met some MiniMax people before, so I left the trip early before MiniMax on this one. But MiniMax was quirky. They have a ton of women in their company, which was very fun. And they have products. They’re maybe slightly more product-focused, but I feel like the quirkiness of the company kind of matches maybe Western confusion over what their products are doing and what they’re trying to do. But it kind of matches their language models that are a bit more efficient.Grace Shao (26:35)Well, they came out with a lot of very consumer-focused applications, right? They had Hailuo and Talkie, all these character companion-bot products before.Nathan Lambert (26:45)Yeah. And then the last one I went to was Ant Ling, which is also very corporate, but in a less intense way, because I think they see it as serving their own products, whereas Alibaba Cloud is like, this is the gold mine we have to win. It’s a much bigger deal for them than Ant Group. But a lot of these things, when you list them — I don’t know, eight to 10 companies — they’re all pretty reasonable with respect to the age of the company and what the company does best. There’s not as much confusion.Grace Shao (27:14)Yeah. And Ant is low-key best at medical chatbots right now, which I guess makes sense because everyone has access to Alipay. And then for seniors, apart from WeChat, it might be the only application they’re using on a regular basis. So it became the default medical consultation app, which is really random, but it’s their niche now. Yeah, I think you’re pretty spot-on. It’s pretty cool that you got those takeaways, even just meeting with them for a couple hours.Nathan Lambert (27:41)I have been reading about them for so long, so a lot of these priors are easy to confirm when they kind of fit with things you have seen. The Chinese showroom culture is so interesting, and also one of the most surprising things to have at software companies. It’s so funny. They’re definitely appealing to Western audiences. Z.ai had poorly translated merch. What was it? Something so — it would be borderline inappropriate translation in the U.S. It was like “ship big, go hard,” or something. Just some really weird translations. And they have live API statistics in their showroom. So Z.ai was like, we’re serving 5.5 trillion tokens a day. All the U.S. companies are so closely watched for when they announce token statistics.I know at least one of these numbers is wrong. It’s something like Fireworks does either 30 or 300 trillion tokens a day — or I meant Together for that one — and then one of Fireworks or Together, and the other one, are like 100 trillion tokens a day. Don’t take these as sourced; go look them up. There were some public announcements recently, but those were the first updates that anyone has on major infra companies in the U.S. Inference is a huge market. You don’t hear anything from Fireworks because they’re just struggling to demand and they’re making bank, because inference is a much better thing to sell than bare metal.Essentially, inference is selling the software implementation to serve tokens more efficiently, and you can just get more margin when you improve the stack for a fixed model. So a model comes out and you host it, and then you can make your stack more and more efficient on that model. You just get more margin and hopefully growing usage. That’s way different than GPUs, where the best case is that you lock in a huge commitment for a long term.Just being able to walk into an office and learn about their API is interesting because they also had geographic distribution, which was like: China was, I don’t know, two-thirds; U.S.A., 20%; and then the last percent was Singapore, Korea, Japan on the Z.ai API. So that’s cool. This is s**t that I always want to know about the companies, and I have no idea. One of the things I always want to know is: how are open models being used outside of the U.S. and China, and has this decades-long process of technological diffusion started to kick in in a way that any company can measure? I don’t think anyone has good data on it yet, but I think it’s obvious that at some point, open models that are cheap to run are going to have some interesting playbook across the globe for the long tail of countries. Maybe I’ll just walk into the front door of a Chinese open-weight company and get my answer.Grace Shao (30:30)But actually, I think the culture of these labs — a lot of them, because they’re run by really young, passionate people — you would feel like they’re a lot less commercialized or less corporate, or at least less sleek. They’re not sophisticated with, you can say, the capital-market side of things, but you can also say that they’re just really naive and open-minded and passionate about the product they’re working on, with less of a corporate guardrail built around them.Nathan Lambert (30:56)Yeah.Grace Shao (30:57)Okay, I want to talk about...Nathan Lambert (30:57)Yeah, go ahead. It’s like one of the people at Z.ai who’s known on X — I don’t know, 9,000 followers — it’s like Lu. She came up and was like, hi, I’m a student, I’m 20. I’m Lu from X. And I was like, that’s hilarious. There was a lot of s**t like that. It was like, oh, okay. I don’t want to call her a kid, but it’s like...Grace Shao (31:06)Yeah, yeah, yeah. And I think the one that runs Moonshot’s developer ecosystem or something is literally a girl fresh out of school, right? And she just posts hilarious memes all day long. There’s no filter on her social media. It’s funny.Okay, we go on these tangents, Nathan. We need to come back on track. Open source, open weight. Why? Why do you think Chinese labs are adopting it or embracing it, however you want to put it, especially after visiting them? Is it because they simply have to, because of what we talked about — they are leaning on each other because of all the constraints they have? Or do you think the philosophical drive is actually bigger in that ecosystem? Or is this a bigger strategic thinking for diffusion in the long run?Nathan Lambert (31:54)I actually don’t feel like it’s that special ideologically. I think it’s easy to say the ideological line when you are doing it. Now you can look at Zuckerberg: he said the ideological line when he was doing it, and then he stopped. I think it’s mostly just that, for one, distributing within the U.S. ecosystem, especially to enterprises, is the highest-value market, and they can’t sign many enterprise deals. And the closest best thing is things like Cursor adopting Kimi’s model. Even if Kimi doesn’t get paid for that, they’re happy. That’s the biggest sign of credibility for them, and they can figure it out in selling tokens or whatever in the future.Practically speaking, one, the only way to influence the U.S. market is by releasing these models. And two, it seems like they don’t feel like they’re losing as much if they release and share things. If the model was closed, they just think they would get less influence, they would be seen less, fewer people would use the model, their actual paid offerings would be adopted less. It just seems almost overwhelmingly obvious, because there are all these benefits and not as obvious of a drawback. There will always be better models, and just keep going. But I think every scientist loves...Grace Shao (33:08)Then why are so many U.S. labs against it, or not willing to?Nathan Lambert (33:12)Because they can make as much money without it. Anthropic and OpenAI make more money by not releasing them. They can just make so much money, so why bother thinking about an open model that doesn’t make money? There are different scales of influence. Same with Google. Google’s making so much money. I think Meta will make a lot of money by having good AI models in their products, if they get their act together. Even Google could release more models. They have so many surfaces other than Gemini that need AI to be commoditized and used, like the cloud and all of this. Meta could release the models. It’s just not worth the effort for some of them. They’re like, we need to do this high revenue target; it’s too much of a pain to go through legal and make it ready to release. Why bother?I don’t know, maybe it’s a little bit of a cynical take, but I think Microsoft and Meta could release their best models openly because they benefit if it’s a commodity layer. But I don’t expect them to, because it’s just kind of like the benefits of focus are so high, and they just kind of see it as something they don’t have to do.Grace Shao (33:56)And it’ll be good for them. But then eventually, we will see some consolidation in the market as well, assuming — because you can’t really have 10 labs in each dominant country right now all exist.Nathan Lambert (34:26)I do expect consolidation. I think this is potentially a subtle cultural point, which is that the U.S. labs are more likely to buy into “we’re special, we need to go fast, keep it closed,” and the Chinese labs are not. There could be something there. That’s also who the decisions funnel up to. I don’t know. I talked to the Alibaba people that make these decisions. I can’t say all the things that they say about them. Some of these were two-on-one and off the record, so I can’t say all these things. But at all the other labs, there is a person that makes the call, I’m guessing. I think those are senior leadership that we’re not talking to. So it’s kind of hard to know exactly what they really think.I definitely expect consolidation. My thing is that I expected it in China faster because the capital markets aren’t as strong as in the U.S., but I don’t have a model for that. I think you can model it, which is: what do you think the revenue growth would be? What do they need to do to raise to keep training bigger models? What is the compute cost? Then you look at the potential raises and think about which country would not be able to do that race first. But also, it’s this wild thing with OpenAI raising $120 billion. Are you kidding me? What is that?Grace Shao (35:47)Yeah, the valuations in the U.S. are not really understandable by anyone else right now. I think in China — so on your point on that, I’ve been writing about this and I think it would make sense for Tencent just to buy out one of the labs. They have the money, they need the capabilities, and frankly, they’ve really been struggling to compete with their LLMs, with all the labs talked about just now. So my...Nathan Lambert (36:05)Their licenses are so bad. They release all these models that have horrible licenses. They’re not that good, and the licenses are just horrible.Grace Shao (36:13)So I feel like it financially makes sense for a company like that to optimize and just buy out a lab. Then the labs can also lean on their distribution, because at the end of the day, how are they going to win consumer mindshare or distribution in China right now when it’s really just dominated by Alibaba, ByteDance, and Tencent? That’s my spiel. But when I spoke to some of the researchers...Nathan Lambert (36:33)I think big companies have a lot of inertia, and the senior leadership has the call, and they can have inertia. I still think Apple just ends up buying some lab for $25 to $50 billion. It’s not the worst thing. Just golden-handcuff the researchers. Some will still quit.Grace Shao (36:43)Yeah. But I think right now they don’t want to. The labs still have a dream. Some of the researchers still have a dream. So when I spoke to a lot of them, they’re like, no, we don’t want to do that. We want to commit to our own frontier research. If I wanted to join one of the big tech companies, I could have. So why would I want to sell? That’s what the researchers think. But to your point, we don’t know what actually the one person or two people at the very top think, especially if they continue to have hurdles with compute access and capital access, which brings me to the question.Nathan Lambert (37:14)It also depends on your view of inference. You can ask your next question. I don’t need to cut you off. It depends on your view of inference. If these agents are just so much inference, I do think it’s going to be an oligopoly-style market, not a monopoly-style market. And what’s the difference financially between two and four or five big companies with great models? Is that actually not sustainable if there’s so much demand? There are a lot of cases where we have two or three, like the cloud, but what’s stopping that from being four?Grace Shao (37:40)I think they will be the infrastructure providers. Yeah, yeah. And they would kind of lean into each of their existing ecosystems or distribution, whatever you want to call it, and serve certain specific models for specific uses. So enterprises can choose what matches their needs the best as well.I do want to bring the conversation to a more contentious topic, which is on distillation and model convergence. You raise the question of whether Chinese models are structurally different. Often we are hearing claims saying a lot of these labs are about three to six months or six to nine months behind U.S. labs. There’s obviously a lot of noise or allegations and accusations from certain U.S. labs saying Chinese labs are distilling them. How do you actually see that accusation or that kind of dynamic?Nathan Lambert (38:36)The biggest unknown that I don’t have an answer to, which actually has a lot of sway, is how much of the Chinese companies are actively trying to hack APIs versus just showing up as a customer and paying. If you’re trying to hack the APIs, normally you get reasoning traces out so that you can create a reasoning foundation that would be similar to the model that you’re trying to do this from. That’s very different than the API standard form, which is just the output of the model, which is a less direct process for learning from.I don’t know the magnitudes. If it’s more just like, I walk up to an Anthropic API and I use it as intended, but I’m making a competitive model, I’m not very sympathetic to Anthropic. They could ban it if they want to. And I think the impacts are kind of a standard practice. You can do it with many different models and so on. The evidence Anthropic provided is not large enough scale where I’m like, this is industry IP theft at mass scale going on 24/7/365. So there’s definitely some gray area to what is actually happening in distillation.That’s why, on the policy side, I try to push people to not call all of it the same thing. Essentially, using any API endpoint to make synthetic data to train your model is some form of distillation, but it’s very different if you’re trying to break this model so that it gives us a different behavior that is hyper-useful for training and not get caught. Those are pretty different actions, and they’re all looped into this common phrase of “distillation” right now. That’s my biggest problem, which is that academic researchers and small companies use distillation extensively as the core of their business and the core of research methods. So if the U.S. government nukes that as a thing that could be done in the AI ecosystem, it’s mostly bad for small players, bad for U.S.-China tensions, and bad for academics. That’s my primary concern.And then trying to get the labs to actually say more. There’s a distillation side and then performance is the other side, which on benchmarks, it does seem like the Chinese labs tend to be six to nine months behind. When it comes to general use, I’ve always found the closed models to be better in ways that are hard to measure. So I go very back and forth on whether the closed models are better. I think we will especially see Anthropic and OpenAI pull ahead on knowledge-work tasks like legal, healthcare, financial services, because I just don’t see the Chinese labs paying for that data. All that data is going to be people that charge hundreds of dollars an hour to annotate and create these environments. So it’s a whole new capital build-out that goes on there right now. It’s going to be billions of dollars if you’re going to buy a billion dollars of data and a billion dollars of compute and a billion dollars of talent to train your model.Grace Shao (41:30)They don’t have the money.Nathan Lambert (41:30)I don’t think they have that. Mercor has some of these evals, and I think there is a bigger gap there. So it’s very interesting. Florian, the guy that helps me, and I disagree on it. It’s this fine line between, yes, the evals — coding and lots of these things, and even random evals that surely the Chinese labs aren’t training on — the open models really are genuinely crazy impressive scores. So I think there’s also a tester’s bias, where I don’t use the open models as much. Maybe it’s hard to ground in my head what I was doing with AI six to nine months ago. I wasn’t even using Claude Code as extensively.I guess the question is, at the end of this year, can I use an open model in something like Claude Code and feel like it works at all? That’s the test on the performance gap, starting in June, June to August, and whether or not that hits. I don’t think the open models have hit that yet. I think it would be way more of a narrative if all the companies spending billions of dollars on Claude are like, oh, we can spend 1% and just use DeepSeek. These CIOs and all the big companies — some companies spend more on tokens for their employees than on headcount. These are normally startups. But they would happily reduce that token cost to 1% expenditure if it really was that similar, because then you could just use 10x the tokens. I don’t expect that to happen. And I expect things like the latest Claude and GPT-5.5. I expect more of these things through the year, and we’ll see if I end up being right. Both are right at the middle of us, as a world, getting more clarity on them. They’re like 18-month-long stories unfolding, and I feel like we’re just in the middle of performance gap and distillation and learning more.Grace Shao (43:25)Yeah, it’s interesting. You mentioned — it helped me recall a conversation I had with other people as well. The point on distillation is that I just had a conversation with your former colleague at Hugging Face, who leads APAC, called Tiejin Wang. He was just saying, look, the distillation accusations don’t really make sense because we’re all distilling off of each other as we speak. I’m learning from you; you learn from me. We’re distilling. It’s so vague of a terminology to just use that to accuse all these various behaviors.So to your point, I think people in the technical world who understand what’s happening actually want more clarity on what is the gray area, what is actually black and white, and what is not appropriate or unethical. That needs, I think, the industry to come together to really put guardrails and rules around.Now, number two on the compute side and the data side. Something anecdotally will be interesting to you is that when I spoke to one of the lab researchers in Beijing, I think in February around Chinese New Year, they were saying, look, they want to get better data, but they can’t because usually a lot of American labs would pay tens of millions, if not even more, like a hundred million dollars, for a set of very obscure or niche datasets, but they would have an exclusivity contract. What the Chinese labs will do is that they will literally wait out the exclusivity contract and then, say two or three months later, pay for it at one-tenth or one-twentieth of the price for that same dataset. So then once they start post-training on that dataset, that’s where the three to six months or six to nine months come in as well. Yeah. On that note, I want to...Nathan Lambert (45:00)Yeah. I think the data industry in the U.S. has two things. One, the lab asks the data vendor, we need this specific type of data. And the data vendor is a network that connects the people to the lab. The other thing is the data vendors know evals that are important, so they try to create good data for hill-climbing on specific evals. That data could be sold to multiple people, but is less expensive because they make it once and expect to eat margin or take margin on it. There could be a pipeline where once OpenAI is at the cutting edge, creates this new thing, they create deep research, then the data industry is like, let’s make things that are a little bit cheaper to sell. So there is time lag in these various things.But I heard the same thing on the ground, where they have a negative view of the data industry. It’s like, quality is bad, we don’t really have access, we do some in-house. That’s a very big difference from today, which is that you have the data companies in the U.S., which is insane.Grace Shao (45:53)Yeah, the American data companies are so mature. It’s its own sophisticated ecosystem.Before we get into data, I actually want to ask you this question. I think recently a lot of the narrative is now saying, look, Anthropic and OpenAI have kind of proven that pre-training scaling laws continue to hold, especially with the recent models. There’s an obvious compute constraint on the China side that we talked about. And then it will likely be even more amplified with the absence of Blackwells in the coming months.So as we move forward in this race, per se, if you have to put it in China versus U.S. in that sense, will we see a wider gap between the performance and benchmarks between the Chinese labs and U.S. labs? As in, will we see the gap going to 12 months, 24 months, as Chinese labs are very, very constrained on compute for pre-training breakthroughs?Nathan Lambert (46:40)I think it’s more of pre-training as a thing that you could actually finish. How big can you pre-train a model that you can finish and serve? The Chinese labs could train models that look like GPT-4.5, which is this giant model, but you can’t serve it. They end up training a model that is 2.5 trillion parameters and they release it, and no one can use it. They could barely serve it on their API because they don’t have Blackwell NVL72 racks or something — these racks that are definitely what are serving these large MoE models. They just don’t have the quantity of these.So there’s a difference between models that you can build and models that are actually useful. I think some of the Chinese labs are definitely like, we don’t need to release the gigantic models because nobody is going to use them in open weight. The biggest models end up getting served via API. So there might be some segmentation in that market. But I do think the inference and amount of economic resources that you have to serve your customers is becoming a thing that dictates what models are built. That’s why I think the gap will continue to rise. All signs point to GPT-5.5 being a bigger model, and I don’t expect that to stop.And then the economics of it is just the basics of: you need a certain volume to have the margin to support the research, because you can’t keep raising these ridiculous rounds forever. I think OpenAI, Anthropic, and Google are the only people with that AI usage volume to keep marching down the scaling laws to another 10x of training compute, which is mind-boggling amounts of investment in a model. That’s why, when the economic markets slow for fundraising, the model gap between these big three will just show a lot more. That’s the distilled way to say my prediction of when things will look different. It’s like these labs can’t fundraise, they go public, they can’t generate revenue more on their paid services, and then it’s just: look at how much training compute can be allocated or can’t be allocated.Grace Shao (48:41)Yeah. Basically, we’ll see a bigger gap, I think, in the coming months. Then what can make up for that? Domestic chips, or, like you said, better data. And why is it that sometimes people assume China has a very strong data ecosystem or data products, but actually the data vendor ecosystem is very weak in China?Nathan Lambert (48:41)So generally, I think I agree with what you said. I don’t know on the data side, but the way domestic chips could help is that if Huawei chips are fine for inference, and if they have sufficient volume to support the inference economics, which then trickles back into revenue, my read is that they just don’t have the volume of the chips, especially spread out across the amount of companies that they have. Essentially, the total FLOPs of Huawei, all the things produced, and it’s going to all these different places — it’s just not big enough.It could be something like ByteDance and Alibaba, with offshore data centers, can keep up a lot longer because they have access to Nvidia compute and have for a long time through this kind of offshoring. Maybe that stabilizes the ecosystem, and we’ll see what the AI startup, the younger startups like Kimi and Z.ai, end up doing. No one wants to do this, but if they pool resources, they last an extra year. You get another order of magnitude if they all pool together, but I don’t see them doing it.Grace Shao (50:00)But that’s the thing we were just talking about, right? MiniMax and Zhipu, how can they possibly compete with the hyperscalers at this point if you need offshore data centers? And the fact that Zhipu is on the Entity List doesn’t help, right? It’s not going to be easy for them to access these data centers either.Nathan Lambert (50:12)Yeah, I think they can’t. I think they won’t. Human nature will make it so they won’t collaborate. They’ll just do something smaller. They’ll just have successful businesses that are different.Grace Shao (50:22)They just have smaller ambitions, want a smaller piece of the pie. Yeah.Okay, so you wrote something like, nothing’s a secret, but everyone wants Nvidia chips. They want it, they don’t know how to get it, they’re fighting over it.Nathan Lambert (50:34)Yeah. They’re the only thing that works for training. All the models are trained on Nvidia. I don’t believe the DeepSeek propaganda that it’s trained on Huawei. The only models that are trained on Huawei are tiny. Inference on Huawei works. Every lab is like, inference on Huawei works. The labs that don’t have meaningful inference are like, we are told to get Huawei, so we buy them, but we don’t use them. Earlier research labs are like, we don’t have any inference and we don’t have a need for Huawei. Any company that has meaningful use of their models has figured out how to run them on Huawei for inference, which, to Jensen’s credit, is like — it’s happening when he said it was going to happen, but it’s not that surprising.Grace Shao (51:11)Yeah, the Dwarkesh interview. I don’t actually understand why he got so much hate for it because even without your political stance, what he said actually made sense logically by saying, if you don’t sell them the crappier versions of what we have, they will have an equally quite crappy version to serve themselves, or they would just want...Nathan Lambert (51:28)I think they would buy both. Buying both is actually true. The amount of Nvidia chips that you would have to sell to China for them to stop buying Huawei — because Huawei is almost surely way cheaper because Nvidia margins are insane — when would they actually stop buying both?Grace Shao (51:43)But then you have to go on CANN. You have to reroute everything back on CANN. The developer ecosystem is not there. That’s Jensen’s point, right? Or the habits are not there. So I think that’s what, when I talked to a research lab...Nathan Lambert (51:50)Yeah. But I’m saying they would also use Huawei. I think they are so supply-limited, they would use both. Anthropic uses everything. A lot of companies in the U.S. will use multi-platform. Meta is a huge buyer of AMD. Demand is so high that any chip that is potentially viable on the models within a few generations is very valuable. And the fact that you can run some reasonably large model on any Huawei chip is a big line crossed for Huawei.I don’t know if they can produce the volume of chips and scale that quickly, especially as they try to move to lower nodes. That’s the standard semi debate. But the question is: can Huawei scale production? That’s the only question. And if Huawei can manage to scale production, Jensen will just look really right. If Huawei can’t scale production, Jensen will look a little bit like a lunatic, but it will be outside of his hands.Grace Shao (52:43)And we don’t really know what happened during this trip. It seemed like nothing really substantial happened after this big Trump delegation. It was more like a high-profile tourism trip versus an actual deal trip.Okay, I want to ask you something you wrote about that’s a bit niche, not something you usually write about. It’s on the SaaS side of things. You said that there’s a common argument that China struggled to monetize AI because they’re unwilling to pay for enterprise software. We looked at how China tries to monetize on consumer AI, but clearly that’s not really been proven yet. In your piece, you push back on the claim and say that there’s a distinction between SaaS spend and cloud or inference spend.Tell us about what you think about that ecosystem and how Chinese AI labs are trying to make money maybe a bit differently from American AI labs.Nathan Lambert (53:32)I don’t know if it’s necessarily different, but I ask a lot of researchers about this. They say that everybody is trying the new AI tools when they come out. If they don’t like them, they stop using them. If they like them, they keep using them on the consumer side. So something like Claude Code would be an example: tons of people tried it. I’m guessing lots of them churn in China, just like in the U.S., but consumers are very quick to adopt and try new things, but won’t stick if it’s not actually serving them.And then the enterprise is like: there’s definitely cloud that exists. Digital services are gigantic. They essentially think that there’s more runway for making money on AI models that falls into that. And they all use coding agents; they all use Claude. It’s a hilarious thing. They’re all very Claude-pilled. There’s almost no mention of Codex, where in the Western media, Claude versus Codex is this whole thing. They all use Claude. And that is obviously a paid service. So I think there are cracks in the argument, and I expect AI models to be seen as a bit of cloud, but potentially it is the thing that changes some of the expectations, where it’s just so transformative because they’re so competitive, and it could be seen as a bit of a phase shift.Grace Shao (54:41)Yeah, and I think it’s a generational shift, a phase shift. Also, actually, recently Doubao raised their prices on Seedance usage and whatnot, and it’s a shift into trying to capture the prosumer market. You can say the average uncle and auntie on the streets still don’t want to pay for a consumer app, but I think there’s more prosumer market share that could be captured in China, maybe not fully enterprise either.I want to ask you about government roles and geopolitics. I know there is a common narrative that usually people assume Chinese AI labs are heavily subsidized. Actually, when I was in San Fran in March, I was at a dinner with a couple of investors, mostly public investors, and one guy asked me, “Hey, are all labs just basically subsidized by the government?” I was like, definitely not. The majority of them are not. If not, they frankly don’t want to take money from the government.It was really hard for him to understand that, because I think the misconception is all Chinese labs or Chinese tech are just funded by the government. Kind of to our point earlier, where any affiliation to any government agency, just by default, is assumed to be therefore backed. First of all, the government, I don’t even know if they have that much money to give out. Number two, I don’t think that’s how competition works, right? So what’s your thought on all of this?Nathan Lambert (55:55)It seemed more like a provincial government trying to help the companies do stuff, which is like get offices, get talent. I don’t know what the provincial government can do. In Beijing, there’s Beijing Academy for AI or whatever, which is a real research institute that’s just funded by a certain neighborhood in Beijing. It was like, okay, the U.S. could do that. But much less of the Ant Group-style thing, which is government takes major ownership stake in an investment round and goes on. Maybe Kimi’s latest round, there were mentions of government-backed VCs, and I don’t know how that kind of intermediary works. So I still think it’s very indirect. And because the government system is so competitive across the different layers, each of those layers are competing to help the companies, but they don’t have piles of cash sitting around to buy GPUs.Grace Shao (56:44)No, they don’t. And they frankly don’t know what they’re doing half the time. This is an argument like what you said: Haidian District or Chaoyang District of Beijing will be funding an academy, and the academy will be in the effort to help AI go toward AGI. But the reality is they’re trying to follow this high-level KPI of being like, let’s make AI happen. All they want to do is write in their report and say, we funded something about AI, so we’ve hit our quota. I don’t think it’s as hands-on as people assume.Nathan Lambert (57:09)Yeah. If you read Breakneck — most U.S. tech people haven’t read Apple in China and Breakneck — and all you need to do is read these books and learn a little bit about the interface between tech and China and understand that they are also hyped about AI, and then you’ll understand that it’s a messy trickle-down process in the Chinese government. It would be very obvious if they were nationalizing a lab. It would be as obvious as if it was in the U.S. It has not happened.Grace Shao (57:18)Both great books.Yeah. So to close, what is the biggest disconnect between how the U.S. AI ecosystem right now thinks of China, Chinese AI, and what you saw on the ground? And what is something you think we didn’t touch on today that you want to share?Nathan Lambert (57:52)This is the thing that everybody asked me. They normally asked me the first thing when I got off the plane: what’s the big thing? And it’s like, I don’t think there’s anything that shocking. I think that many people just haven’t read basic books about how tech is interfaced with the government, and know these things, or hear narratives that are very geopolitical, which is targeting the top end of the government system and how that in the U.S. engages. And there’s a lot of shrapnel from that. Anthropic pushes very aggressive China narratives, and Anthropic is a very followed company in tech.Most people don’t spend the time on this in the U.S. ecosystem and just don’t go deep on it. I don’t have anything shocking. It’s good to encourage people to do some of that because these dynamics impact things like: the Chinese open models are really influential, and now Silicon Valley is building AI. So it matters to a lot of people, but they don’t study the causes of why they might do this. They just are like, it’s here, I don’t need to think about China.Grace Shao (59:00)Yeah, and I don’t know what it is. You and Bill Gurley were saying this in a couple of his public appearances. It seems like Chinese researchers, tech people, CEOs, whoever, are a lot more aware of or following more closely U.S. leaders and thought leaders, tech leaders, business leaders, than vice versa. There’s something about that. I don’t know if it’s just easier to dismiss it or easier to not have to learn something new. But the goal of AI...Nathan Lambert (1:00:27)I think it’s American culture. American culture is very obsessed with its own weird world. Yeah, it’s hilarious. American culture is ridiculous. It’s so ridiculous.Grace Shao (1:00:27)As a Canadian, I can’t say things like that. You said it. I’ve lived in the States, but I can’t say this. But I think, look, shameless self-plug here: as a Chinese Canadian, my life goal here with AI Pro is really just to bridge that gap. I think to your point, there’s going to be geopolitical narratives and rhetoric at the very top, but for the average person, or even for builders, tech people, whatnot, it’s probably in everyone’s benefit to understand what’s happening on the other side and stop alienating it or stop making it as if it’s so different.I think throughout this conversation, it’s really just to say, look, so much of it is so similar, but so much of it is slightly different. The difference is not really a government mandate versus maybe a cultural difference or resource-constraint difference, especially in building technology. But that’s kind of my view.One last question for you, which is a question I ask everyone on the show. What is one differentiated view you hold? Throw me something crazy.Nathan Lambert (1:01:28)I know, I’ve always kind of been open-models doomer, even though I build on them. It’s just that it’s so unsustainable, and there’s so much money to be made with building closed software, that I’m constantly doomy about the prospects of open models. I’m always a skeptic.Grace Shao (1:01:40)It is a bit sad, isn’t it? How does a company like Hugging Face actually make money?Nathan Lambert (1:01:45)I don’t know. You can look into how much money they actually make. It’s not very much, unfortunately.Grace Shao (1:01:48)Yeah, I think that’s the unfortunate reality of the capitalist world we live in. As much as it incentivizes the competition and breakthroughs, that doesn’t help with what we just talked about earlier. Yeah.All right, Nathan. Thank you so much for your time. Really appreciate your insights and your sharing.Nathan Lambert (1:02:04)Yeah, thanks for having me. Good to see you.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  AI x education, a contentious but unavoidable future. Designing tech for children with Dex's Reni Cao 18.05.2026 1ώ

  I spoke with Reni Cao, the CEO and co-founder of Dex. Dex Camera is a language-learning camera for kids. Reni is a dad, a former product lead at YouTube, and on a mission to build technology that does good for kids and gives digital autonomy back to parents. We dive into his personal story from his high school days that drives his passion for AI, and why he believes the current education system is a “cookie-cutter” that fails curious kids.We get really into the nitty-gritty of what makes “good” tech versus “bad” tech for kids and why the category of ‘children-first tech’ is very overlooked. Reni explains why most children’s apps are built on an “attention economy” model that forces them to compete with addictive content, and why his team needed to build physical hardware to break that cycle.We tackle the hard questions, including the pushback from parents who believe in “no tech” childhoods. And he shared his most non-consensus view: that the era of standardized, industrial education is over. He believes we are entering a golden age of “scaled homeschooling” where AI meets kids where they are. Whether you’re a tech investor or an anxious parent, this conversation about nature versus nurture, “nei juan” (involution), and raising resilient humans in an AI world is a must-listen.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently. Season two will host a series of guests from early-stage investing, as well as builders, founders, and product managers.For more information on the podcast series, see here.To find the previous episodes of Differentiated Understanding, see here.Chapters00:00 Reni’s Journey to Dex Camera03:48 Designing for Children: Principles and Insights08:05 Technology’s Impact on Child Development12:09 Bridging the Gap: Business and Product Design15:36 The Role of Parents in Tech Development25:20 Leveraging AI and Language Models29:48 Value-Driven Pricing Strategy32:05 Defining the Product Category34:33 Subscription Models and Content Delivery37:58 AI and Parenting: Balancing Technology and Safety43:29 Unexpected Use Cases and Impact47:29 Personalized Education and Parenting PhilosophyAI-generated TranscriptGrace Shao (00:00)Reni let’s start with your personal story. Who are you and who are your team members? Because when I met you in SF, I was so enamored by the product and I thought your story was so interesting. So please share that.Reni Cao (00:11)Hi everyone, my name is Renny, CEO and co-founder of Dex. We’re a technology company in San Francisco, almost all parent company, which is pretty special in a startup setting. We’re a bunch of parents that having trouble with the same kind of like a reality where like our education system is a sort of like cookie cutter and our entertainment is also cookie cutter for children.So we’re like, can we harness technology, especially the latest development of the AI, in different way for families that really gives children a chance to become the best version of themselves and ⁓ give the digital autonomy back to parents themselves rather than accepting the fact that they have to struggle between technology versus no technology. So yeah, we’re the parents, of like a bunch of missionaries in this journey together to explore how can we make the best use.of the AI and our first product is called Dexta Language Learning Camera where kids can take pictures and turn the whole world into language immersions. And it’s a product targeting young children three to eight. And we’ve sold 10,000 pieces so far and ⁓ ratings has been high and we’re pretty excited about this. But yeah, this is pretty much about us.Grace Shao (01:24)But Reni, tell us a bit about what you did before Dex actually. What kind of led you to this path? I know becoming a parent really did inspire you. You have a young daughter, I think similar age to mine, around three years old. But before that, what really led you to this path? Were you always passionate about children’s tech or education?Reni Cao (01:41)I actually have been a product management guy for the last decade in Silicon Valley, some big companies like YouTube and LinkedIn, some smaller s***, ZFS, Wish. But I have been a builder since the beginning. I would actually say that my passion for decks actually originated much earlier than I started my career. It actually started right when I was at school, but happy to say more if you’re interested.Grace Shao (02:07)Yeah, no, do tell us a personal story there.Reni Cao (02:09)So I was always this random kid with tons of questions back in high school. And very unfortunately, I think the education system, especially in East Asian countries, is not designed for meet kids where they are. So every time when I come up with a random question, my teachers are usually a little bit impatient and will be like, can you just go back and finish your quiz, et cetera, et cetera.So the moment I saw when GPT-4 comes out, I was thrilled and I posted a long like blurb on LinkedIn. Basically saying like, you know, if I had, have this as a kid, I would have grown into a more complete human. So this kind of like, I feel like this like generative AI’s capability to meet kids where they are, especially meets your needs for curiosity. It’s game changing. So.I feel like I’m building this product first and foremost for a younger me that could have benefited so much from this. That’s pretty much the story about me. yeah, I know we see it and of course our parents right now we see there is a tectonic shift in terms of the skill landscape and what the future of workforce is going to be and even the existential challenge of what does human mean in a future society.So we do want to build something that’s centered around children, centered around the family to help them find what they love and build agencies around it at the end of the day. So yeah, that’s the two main driving force of me coming to Dex. But I would be honest about it. It’s like very random. When I want to start a company, a lot of my colleagues are very surprised, being like, oh my god, Renny, you’re getting into this field. But yeah, I guess I finally find the work of my life.Grace Shao (03:48)I love it. think you need to understand the passion and the personal reason behind the businesses to really understand why the design was frankly so intuitive and why you’re so passionate about building this and leaving such a comfy, know, like cushy corporate role. I think that’s the one thing that stuck out to me. The product itself is actually so natural to how children behave to your point, like my three year old.from morning to night, know, morning she wakes up, it’s like, mommy, what’s this? What’s this? What’s this? What’s this? How do say this? Why do you know that? Sometimes she gets angry at me. If I don’t know something, she’d be like, but you’re an adult, you should know everything. But the reality, especially with languages, it’s really difficult. So for example, yesterday she was coming back from her Mandarin class and she said, liu shu, she was pointing at random tree. And I was like, that’s not liu shu. All I know is not liu shu, but I actually don’t know what liu shu is in English because I think it’s only really common in mainland.I’ve never seen that kind of tree. Well, I guess it’s a willow tree. You don’t see it very commonly elsewhere. And then she kept on pointing at trees, but in Hong Kong, you clearly don’t have liu shu because Hong Kong is like tropical. And then she got really, really mad at me. And that moment I was like, wow, if we had a Dex camera, that would have been perfect. But I was literally trying to take a picture of it while we’re moving car and try to upload it to GBTB, like what tree is this? What’s the name of it? So anyway, I think it’s really great product design. And I want to kind of get into that a little bit.When you were designing it, what was the thinking? Like, what does it mean to be children first?Reni Cao (05:10)I think there are three layers of children first as a principle. The first layer we already touched upon that. So young children, their hand anxiety is very different from adults.they tend to use one hand to operate a device and another hand they want to use for sensory explorations, like they want to touch. Sometimes they want to just move things around. So this requires a different form factor that one handed use, very tactile, very intuitive for young children such that they can explore a world while harnessing the power of AI in this case. So this is kind of like the user, the special things about the user.And it’s a different design. think that’s layer number one. I think the layer number two is also that the device itself is a metaphor for the market as well. And in the market, we want to build something that’s drastically from the so-called adult-centric smart devices, namely the phones and tablets, to send the market a message that there could be a different option. There could be a good technology. There could be a family-centric technology. And we’ve picked this form factorutilizing the metaphor of magnifying glass. It is something you use to see some hidden wonders, otherwise you cannot see. I do think that’s the ⁓ second layer of the things, which is like metaphor and category creation. And at the end of the day, I do think we intentionally make the device kind of worth finding this fine balance between engagement andlearning or kind of like a healthy aspect of the technology, meaning like we add a assistive screen, but we make it really kind of like limited and not the center of the whole kind of like a user journey. And we want to kind of like find a new way to put all the components in our consumer electronics world in a way that it strikes a more delicate balance and ⁓ let the device itself to be kind of like, you know, retentive.for children without getting them to be addicted. So it’s kind of like we intentionally make it a little less stimulating, actually much less stimulating than a lot of a thought-centric ⁓ product. So that’s the main three kind of like principles around the product design. There’s a lot of conflicting constraints here, as you can see, but we do our best trying to find what is the answer. And here we go. Like what you see right now is our first, you know,⁓ answer we have thought through and I think the market validated the answer quite well so far.Grace Shao (07:39)Yeah,definitely. think exactly to your point, know, like a lot of times, I think when we as young parents looking at introducing technology to children, really worried about the big screens, addictive nature, or even the parental, even though a lot of them allow parental control, it’s the unlimited access to a wild, wild internet out there. Like all of these things are basically concerns and or reasons why we hold back technology from our kids.So actually on that note, do think you kind of mentioned it, right? Like technology over the years, especially big tech frankly, has garnered a bit of a bad reputation. And I think that was really tied to the rise of social media and all of this mental illness that came with it. And obviously like you mentioned the addictive nature. So what do you think is actually harmful to the children’s development when we are looking at tech? What are areas actually we can really embrace technology?I think you kind of touched on it lightly, maybe explain it to us in an even deeper, more technical way.Reni Cao (08:37)Yeah, our thesis is that why a lot of parents think technology is negative for a good reason. And the reason is that all the main status quo technology for children are built on top of the attention economy, as we call it. Everything revolves around time spent and how much attention, how much engagement in terms of like a minute, seconds, sessions you can get.That, is the reality because, think about it, you build an app on an iPad, immediately you’re entering a competition with Roblox, with YouTube Kids, with all the videos, all sorts of things out there. You could do well. You can try to do good for the society, for the families, but you’re effectively competing against more like...addictive kind of like a form factor of information and it’s a losing battle and as we call it is a rat race. So no matter what type of like educational apps or content you’re trying to deliver at the end of day you have to deliver them in more and more engaging way more and more gamified and more and more animation used etc etc. That’s I think that’s why it’s another reason why we need hardware at the end of day. I think the first stephow we can create alternative reality is that we need to create a new world, a new kingdom where the business is built upon outcome rather than attention. Meaning like it’s not the time spent logic anymore. It’s like, can you use this device? For example, for Dex, you can use the device and you see the child speaks better after two months or your kid starts to have a like a love to speak Mandarin and not preserve the rest of their childhood.I do think there is a business model there like that, but I believe that business model warrant a totally kind of like a different design of the experience from ground up, from the device layer to the software, to the content, all the way to like user interaction. So I do think like that’s why the current technology is considered bad because it raised towards attention. And I think ultimately, inside Dex,I believe the final answer to create that alternates like a reality is can we deliver something that’s purpose built for children before we build a general sort of like, you know, like time spent logic, like a product in, in, in the case of the decks, is something that, you know, purpose built around the languages. cannot do a lot of things. It cannot, it’s not a chatbot. It cannot, it cannot play videos.But I do think even do one thing super well with the Frontier technology already delivers so much value to the families such that you can build a viable business model on top of that while creating values for families. I think being courageous enough to limit our scope to something to begin with, like really hold onto our principle, deliver a promise, create values there.is another internally operating principle to get there in terms of how to harness the technology. And I want to say that it’s very interesting. What we noticed that is a lot of people are trying to use the AI in quite an all-in-one way. So you can see a little device with tons of features in there. can generate pictures. You can talk to celebrities at chatbot. You can talk to Elon Musk on that device. And we think, actually, that would be a very slippery slope.⁓ in terms of harnessing the technology at the end of the day. yeah, purpose-built is another very critical principle we’re holding on to, to create a good technology.Grace Shao (12:09)No, I love that. But I mean, from a business perspective, sometimes people might not have purpose built businesses, right? Unfortunately, some are not. Then thus, how do we basically help the industry align the business incentive to the product design incentive? Because, know, like what you’re saying right now, it makes a lot of sense. And I think once I saw Dex Camera, I was like, wow, why is there not something like this on the market?but it does feel like there’s a huge gap where, like you said, there is a big devices and the big tech. There’s this tiny niche little products, whether software product or hardware for children’s ⁓ use, but it doesn’t feel like people are taking it seriously, even though we all know parents are willing to spend on children if it’s for their good. It’s not like the economics doesn’t make sense. So why is there’s that gap right now?Reni Cao (12:56)I think you’re hitting on one of our most recent realization that the parenting needs and the children’s needs are quite long tail or as we call it, very like a versatile, right? Different parents have different parenting needs. Even when you look at the language as example, there are tons of different languages and even more dialects you wanna learn. Like let’s say you wanna learn Mandarin, you still got so many like a dialects there. There hasn’t been a real...kind of like technology that can enable a venture scale business that attracts talent, that attracts a good backing in terms of like a capital to build something that’s like a generational. But I do think this is the moment AI is strong. We finally have to make sure we can build one system.that can consolidate all those long tailed needs. Even for Dex, very specifically, you can learn a lot of languages and even more dialects with just like a nine person team building the hardware plus software. I think it’s the catalyst that’s much bigger than Dex itself. And I’m really excited about that. But I think another very interesting angle is like, despite the technologies there, you have another question. It’s like why there is notmore company like Dex. I have a personal opinion here. When new technology comes out, people will tend to use it in the most sloppiest way possible. They were trying to just like, OK, you can chat with the AI, so why don’t we just shovel AI into a little box and put it into a Talking Fluffy and call it an AI toy. And that’s it. That’s my business. I do think it is like a gravity that’s pulling people away.from deeply think how to harness technology and pulling them towards something that’s so trivial and it’s just almost like a shortcut. I think that’s kind of like also, I would call that a trap on the entrepreneur side, that the technology is changing so fast and everyone’s a full mowing, everyone just wanna use it in some way. But I think in this sense, we as Dex, the company, we believe in that.we need to think very deep about how should we use this technology to meet users where they are and deploy like AI in certain ways so Shell can deliver the value. So that’s why we start small, but we’re going to expand from there.Grace Shao (15:09)Yeah.No, it makes a lot of sense, but I think I wonder if you guys all being parents like you just said have made a huge difference. I hate to overgeneralize, but like, you I’ve been in the tech space for 10 years, but usually either I meet men who are like 20 years older than me or they’re very young men who have not, you know, settled into a family yet. And I’m just saying when I tell people my mom, it scares people. They’re like, I don’t know what to say. I’m like, OK, like.I’m not trying to scare you off by telling my mother, but the reality is most of us one day will all have families. And when we do, we start thinking about the things around us very differently, our perspectives shift. And I think to your point when you guys had a lot of purpose designing this product, I wonder if it has a lot of, you know, reason because you guys are parents. Whereas if someone is an entrepreneur for the sake of being a business person, they might not have the nuanced understanding of what a kid needs and what they even think is good for a kid.So to your point, they create little stuffed animals with an L-I unplugged into it, which is horrendously scary. I would never introduce that to my kid, right? I’m getting very, very agitated about this. But you know, another one that we talked about kind of offline was like, I should be ambassador and be paid by Tony Box at this point, because I probably gifted at least like 20 of them out to friends with kids. I think they’re just like, on the surface, you think about it, they’re like, ⁓ a little box that plays music. You’re like, this is so easy. I can just use my iPhone.Reni Cao (16:13)Me neither.Grace Shao (16:32)to exactly to your point. It gives the kids agency, allows the kids to start navigating the world themselves and have preferences. For context for people who don’t have Tony boxes or kids at this point is you put these little miniature IPs, essentially they’re Disney or whatnot, and you can put them on the little box as a magnet. And then the box starts singing and has like seven or eight pre-programmed music or ⁓ stories. And then you can control with your little hands. And basically like you press theReni Cao (16:54)stories.Grace Shao (16:58)big ear, the ear just like the volume goes up, small ear, the volume goes down. It’s like really, really great. So basically introduce technology to kids where they’re like, oh mom, I can control what I want to listen to today. But I don’t need to nag you about it to control the iPhone. I don’t get exposed to a screen. And I can sit there and be entertained for like half an hour myself. So I think Dext really falls into that category for me. Like, you know, we kind of skip the part where we explain how your technology work really and in a very day to day way.It’s basically like you hold a camera, you point at things, you click the button, you can say, what is this? And you default choose languages, right? You actually explain better than me, please.Reni Cao (17:35)So there are actually four questions here. So I want to actually react to all of them one by one. I think this is a lot of good insights here. I think Tony Box and Dex share one thing in common, which is they are children-led, or they are child-led in this case. Think in the POV of a child. The world is kind of like a scary place that you’re told to do this or that.you are brought to here or there, there’s not much quote unquote autonomy you could have. But now there’s a device that your parents actually are willing to let you operate and you can decide what type of content media or interactions you can get. That is just a huge reward to children’s like unlimited curiosity and their like a strong needs to be considered sort of like, you know, a big kid or aeven grown up in a way. I think that’s the intricate magic that if you were not a parent, you haven’t interacted with children a lot, you will miss. So instead of saying like a parent’s made us a better product builder, I think at the end of the day, it goes back to the product 101 that you really need to know your user. You really need to know who are using your product. We spent such a long time with our kids every day.And early days, which is very funny, like ⁓ the first group of users using DAX is just our own children. And that gives us a huge edge there. Right. And I do think you mentioned that a lot of like startup founders in this category, sometimes they’re doing something with raised eyebrows of the parents. I do think they’re a little bit distant from the kids is one reason. And another reason is I do think there is a misconception that children are less.at the end of the day, lot of founders think, you know, those are toys or some gimmicky stuff. Kids, you know, you just give them something that can flash, they can make some sound, and children would love to use them. But I reject that answer. I think that assumption is completely wrong. Children are actually smarter than adults in certain ways. They just cannot verbalize it. But as I said, they already got their little taste.as the famous word, popular words, they got their taste and they sometimes can tell what’s a soulful piece of story versus it’s a very sloppy kind of story. So children actually knows that and they want quality experience, they want quality product, they can actually absorb something that’s really built well for them. I think that just gives us kind of like this endless.sort of motivation to polish our product as if we’re building this for the most critical sets of adult users because we think actually children are more and they deserve more. Now, coming back to how Dex works at the end of the day, I think the core loop of Dex is quite simple. You just take the little camera. I’m happy to actually send a video to be the bureau here. You just take a picture.⁓ And they would just literally just tell you, let me actually take a selfie here. Hi. Let’s see what I can learn about this. Look at that big smile. It’s like spreading happiness everywhere. Can you say a smile?Just smile.smile.Yeah. This is like you get unlimited, like smile comes with some laughing too. It’s when you make happy sounds like, ha. Can you say laughing? Laughing.like the ones we use to listen to music. Do you like music too? Can you say headphones?This is actually English immersive mode. So you can, you can improve your vocabulary there.Grace Shao (21:06)how many languages you have now.Reni Cao (21:08)We have 16 languages and more than 30 dialects and it’s still expanding. And interesting observation here is like the smaller, the more niche the languages is, the stronger the demand is there, which we find is super interesting.Grace Shao (21:21)probably just harder to find offline solutions otherwise, right? Or like harder with the communities, assuming you’re an SF, finding a Mandarin community is not that difficult. You know, if you’re in England, finding a French community, probably not as difficult. if you go, you were saying like maybe like Arabic languages like that are not as mainstream, maybe in San Fran, you have people in San Fran wanting to do that, right? Or like people in Dallas last time you said, trying to learn Mandarin, which again, you don’t have a huge community. Very interesting.I’m sorry, I got very passionate about the topic. So I want to of swerve back to our conversation here about raising children with technology. I’m sure you get pushback. think people right now, there’s the other side of argument where everything should be organic. Everything should be very simple.Reni Cao (21:53)Yeah, of course.Grace Shao (22:08)And I myself, I’m a big fan of a lot of the Montessori toys. You know, they’re not buttons or not even power charged. They’re just little wooden blocks, but they’re designed very well for them to, you know, develop motor skills. So how do you kind of explain to parents today who are saying technology should be rejected in the childhood. Kids should just be reading physical books. should learn the way that we learned or even like previous generation learned. We should go back to touching grass only. SoLike, yeah, what’s your argument there?Reni Cao (22:37)First of all, you are completely right. Every once in a while, we got a comment on our social media that, why don’t you talk to your own daughter to teach that language? Why do you need a device to do that? So your assumption is completely right. And my response to that is, first of all, actually, I respect that parent a lot. I believe in the most ideal world, organic human-to-human interaction and free play in the real world is great. There’s a lot of tech, like researchers actuallyProve that right, right? However, I do think the parent miss out constraints here. Number one, you may want to talk to your daughter, but you don’t know Cantonese, for example. So there’s no way for you to teach some subjects or some skills that you want them to learn or you want to immerse them with. And second, all of us know that the contemporary society is more and more fast paced. Not all the parents enjoy this privilege.of saying, let’s slow down, set up a dedicated time for children to go out to places. All sorts of this ideal family style back in the 80s and 90s changed a lot, I would say. So we are, believe, rather than just blaming the parents, not spending enough organic time with their children, I do believe that technology should be introduced more as an option, as kind of like a gap stop.as one of the extra tools on the table. That’s why when we design decks, we don’t introduce chatbots, but we spend so much time on sharing the insights that what your children are interested in. What did they take a picture of? What do they want to geek on? What did they learn today towards the parent app? And just give them this little window to see the world through their children’s eyes. Give them good downtime topic.giving them a way to reconnect even as asynchronous. So I do think the concern is real and the overall kind of like, you know, judgment is very well reasoned. But I think what that’s the approach here is much more nuanced than saying like, let’s use technology to replace human. It’s not, it’s actually using technology to connect the humans, connect the parents and kids better. That’s the nuance I have to take a bit.Grace Shao (24:42)I see what youYeah. No, no, I love it because actually I’ve seen some parents even give kids like little Kodak cameras these days and these little toddlers go around the world, take pictures of how they see the world and they’re so cute. My own daughter sometimes takes my phone and takes pictures around the home and I come back with a lot of selfies and pictures of her sister’s foot or it’s just very cute because you see the world through their eyes, right? And it gives like, it’s like technology doesn’t take all connection away.on technology. wanted to ask you about the technology. How do we understand that? Like how are you actually leveraging LLMs? How do you route through different LLMs or different languages? Is this something we talked about briefly? But I wanted to understand that bit more.Reni Cao (25:20)to share details. Where should we start?Grace Shao (25:22)Like how does it work? right now? So basically for the little Dex camera, can’t ask it, like he’s to your point, you didn’t build a chatbot. So I can’t ask a question. I can’t have a conversation. It’s not a companion, but I can ask it what’s this? How does all that work in terms of the back end technology and the guardrails you built up?Reni Cao (25:38)Yeah, I thinkin a 30K feed view, Dex are utilizing basically all the multimodal LM capabilities to understand what the children are looking at. And on top of that, we build sort of like a profile, interest profile for the children and the parenting need profile for the parents to help contextualize, you what responses should we give in that case? To give an example, if you’re a three year old,just starting to learn Cantonese and you are sort of like interested in a bunch of like a museum topics or you love like dinosaur skeletons and stuff like that, we will render you more challenges around kind of like hey let’s bring Dex to a museum and learn about different terms there and it will be English the primary languages teaching entry-level Cantonese things there. So basically like the visual understanding you certainly use like a multimodal LLMThe response definitely use kind of a conversation API of a lot of like an ALM. And I think building out this context layer or this memory layer of like a children’s interest and parenting needs, that actually is more complex. That takes kind of like a full agent system to try to understand what matters, like condensing or distill insights into a profile and gradually kind of injecting that into our responses. I think that’s on a very high level. That’s it.We do use a wide range of LLM, mostly with Gemini and OpenAI. yeah, that’s kind of like the high levels.Grace Shao (27:08)I’m going ask a question you might not like, but I’m going to put you on the spot. When we talked last time, said specifically on Cantonese and Mandarin, you do use different LMS, but the accents can be quite funny. Like they’re a bit off. They’re not native sounding. Why is that? And how do you overcome something like that? Or other maybe non-English languages. Yeah.Reni Cao (27:12)No, ask me.First of all, you need to try again because we have a solution already. But definitely, hit. We are already squeezing. I’m so hard that we’re hitting the boundary of a lot of like, in this case, it’s a TTS of the leading providers. Because I think about it, I’m pretty sure you’re using English plus Cantonese. It’s basically using English to learn Cantonese. Is that the case?Grace Shao (27:50)Yes.Reni Cao (27:51)is a mixture of languages cases. The challenge there is that without fine tuning, there is very limited sample of someone that speaks very good English and very good Cantonese, and they mix them in like one sentences. So the data, the training data to begin with is a little flawed. Either you have accent English or Cantonese as the more common cases. That’s the fundamental root causes of this. And we’re having kind of like heavy lifting tasks to kind of like solve that.And with the foundational model getting better and better, think one day we’ll get there. And we can see that to be fully fleshed out in the next six months. You definitely hold us accountable. And I think this is right observation for mixed languages. It’s really hard. Yeah.Grace Shao (28:32)Yeah, I bet. how does it actually work right now? Like in terms of economics, like people pay you about $249, right? That’s the price of the product pre-tax. That’s not cheap. Like it’s much more expensive than a toy, but obviously bit cheaper than an iPad. How do I understand the pricing decision there and price? And then how does that relate to, I guess, how you pay for your token usage right now? Does that cover it?Reni Cao (28:57)Yeah.Yeah. Oh, big time. We actually have a pretty healthy margin and the tokens are getting incredibly cheap. Much cheaper than where we started. I’m talking about like in 96, 97. It were a fraction of the token cost of where compared to when we just getting started, which is back in 2024 February. At that time we don’t even have GBD4, we have GBD3.5. that’s the kind of like, that’s the kind of like, actually that time we have GBD4 butis we don’t have GPT-4.0. So it’s very expensive at that time. So now pricing. Actually, I have a let’s talk about the user-centric view and a business-centric view. On the user side, we’re actually adopting this value-based pricing model, which is like any enough day, language is a high value skill to acquire. I sent my daughter to a language immersion in the US. I’m very embarrassed to mention how much I spent on that school.Grace Shao (29:33)Okay.Reni Cao (29:49)And if DAX can offer 1 % lift or enhancement on top of that school, the price is fully adjusted and much more than that. So this is what I mean by like, and very funny that you mentioned toy, right? Toy is something that you get it, you play it for a couple of days, then you don’t see it, you don’t worry about it. And this is not what we’re trying to do. What we’re trying to do is we want to use a relatively high price to keep ourself honest aboutthe value we’re delivering to the parent. Do we really teach a language or do we really get the kids to fall in love speaking that language? If we do so, that price is well-justed. If not, we’re going to give you 90 days of free return period. No question asked, just return it to us. I do want to use this pricing model to push us to deliver more value for the user. So that’s one aspect of it. And on the business side, very funny, you mentioned, I hate when people box us.into toy category. I don’t blame them. Natural reaction, but I want to send a signal to the market that if a team of talented people, hardworking parents, put their heart and soul in building a purpose-built device that harnesses AI and delivers concrete results, we could get out from the typical, stereotypical, like a toy average order value band and go much higher. Above that, it’s less about, I to keep myis more kind of like, want to send a signal to prove that the market, we have enough parents waiting anxiously for something similar to this and want to pay a perceptually higher price for it, a premium for it. But yeah, that’s kind of like we landed on that price. And it’s so funny that so many people in the early days tell us, you’re going to do $1.99, because anything that started with a oneGrace Shao (31:22)Premium, yes.Reni Cao (31:36)is night and day different than like two, that it started with two. But I actually, I’m like launching a suicidal mission. was like, let’s actually make it start with two, but let’s deliver more value there because it’s never like, it’s not a retail business at the end of the day. We’re trying to create a new paradigm of digital parenthood and childhood. We need to hold a high bar for ourselves. And the price is very telling, like in that case.Grace Shao (31:59)No, I actually agreeand I think would you categorize yourself in the same box as Tony box vertical? Would you?Reni Cao (32:06)Not really. ⁓ Tony Box is a, I would say they are a content business. they are, same thing with Yoto. Actually, their founders have deep backgrounds in labels, music labels specifically, and IPs. So they are effectively a distribution business that they are creating a new channel to distributing those IPs from Disney, from Spin Master, and et cetera, et cetera. And the other side, you can see that at Dex, we’re notGrace Shao (32:19)I see.Reni Cao (32:31)I think like IP partnership or putting characters on our device. And we actually optimize for value and outcomes, like I promised to you in one of our principle. So I would put ourselves in, I don’t know, the de facto smart device for families. Just very honestly, the family device, the family technology, maybe like this is where we’re trying to go to, but it’s a completely like non-existent category before we’re still exploring.Grace Shao (32:33)Yeah.Yeah, yeah.Okay, like family tech device.Reni Cao (32:59)and it may change how I call it.Grace Shao (33:00)think there’s some moresimilar things maybe in East Asia because the audio learning like you know even when I was very young like I remember my grandma had a 步步高步伏机 I don’t know if you know what that is it’s like those like tiny little yeah yeah basically what it is it’s like people learn English with it and I think it’s very very like mainstream in China for a while but like you know these things been around I think in East Asia because everyone is using it to literally learn EnglishReni Cao (33:11)The steps are fine.Grace Shao (33:24)But it’s very one dimensional. It’s like one language to one language. They basically embed a dictionary, make the dictionary into a digital one. And you can ask search questions. You can ask what this word is. might, more advanced one might be even like with images, but I think, I don’t know, in the 90s, I didn’t see any images. But yeah, it does remind me of that technology and that vertical. haven’t seen something like that too mainstream in the West growing up, you know?I think if I was when I was learning French and German growing up, that would have been so helpful to your point. But yeah, so I want to bring it back to sorry, I just want to bring it back to the the business. On the Tony box comment, I do believe their business actually could be really high margin because their product is only say like 199 or something like that, right? Like they’re the box. But each character is not a 20 bucks or 30 bucks.⁓ My daughter is drying me up here because every two months she asks for a new figure. But my point is, it’s a great business, right? Like that thing just keeps selling. It’s like Spotify and a physical thing. So would you guys have add-on any services, software, hardware, anything?Reni Cao (34:32)We do.That’s a lot of investor has been pushing us regarding this razor razor blade business model. I think for us though, what we are ultimately delivering is a business more like an app store.It’s like where you can get personalized content and software for your parenting needs and for your children’s growth needs at the end of the day.We’re launching, not we’re launching, we launched two tiers of subscription so far to validate that. One tier, $10 per month, you got unlimited LTE, plus you actually got a curriculum packed in like a content library. Every day we give you one topic and in the topic you can explore a lot of new vocabulary, expression, know, new languages and it’s good kind of like content to consume. And I think what’s most interesting is our future vision is actually a $20 per month tier.In that tier, you can actually create activities for your children, personalize. Grace, can be like, I run this podcast. I’m a podcast host. How do I explain that to my kid and make it a little bit fun, exciting, and even adventurous as if the recording a podcast is a little journey? And by the way,my kid likes this way of storytelling. You could give a lot of like a prompt there. They’re actually based on the profile, the context layer, we’re gonna build sort of like interactive, like a content that involves taking pictures, speaking, and just like looking at the device for explaining what does podcasting mean. And this tier actually got really good like attraction. And when we look at their subscription retention,is above like 90 % in three months that shows early signs of product market fit. But this is what I mean by like our business setting of day. We are a channel to deliver like harnessed intelligence to parents such that they can build whatever content and software that adapt to their needs rather than just a purely search, then filter or control kind of like a timer. I really want the digital world to revolve around them, running out of way around. So in this case,Put it in a simple way, we give them a tool to build whatever they want, and we charge on the usage of the tool, pretty much.Grace Shao (36:41)No, I actually really see that. I love it. Because I think my husband was trying to use chat GPT for a while to create stories with my daughter. Like, add a pig, add a dog, add a whatever in this. And obviously, it’s not made naturally for this. So the stories don’t come out as, I guess, natively understandable for children. So I see where this can go. And the funny thing, you use my profession as an example.Reni Cao (36:49)Exactly.Grace Shao (37:05)example, like my daughter just thinks I talk all day, that’s my job, and she thinks that her dad sits at a computer and press buttons all day. So between the two of us, none of us are doing it much, just talking and pressing buttons. So it’d be really great if, you know, I can, I guess, lean on technology to find a better way to explain to children modern day careers, you know, that may be not as easy to explain as, know, mommy’s a doctor, and doctors go help people and save lives, which is like what my family has.you know, explained to us when we were growing up, you it was very clear. I want to kind of go on a little bit more about AI and parenting. I think there’s a huge discourse right now in the US, especially, I think from my point of view, where I sit in Hong Kong, in Asia, even yesterday, I was speaking to someone from South Korea, venture capitalist, they’re saying that parents and society seems to be a lot more open to bring technology into their day to day lives.They’re much more open to the idea of leaning into technology for personal use and less worried about privacy and you know these kind of issues I guess. So at a high level, what do you think, should we be concerned when we introduce technology to children? will they, you know, for example, taking pictures themselves that automatically goes into one of the LLMs. Is that something that...he should be mindful of or are there guardrails that can be built in?Reni Cao (38:26)We should be definitely mindful. That’s why we enforce ZDR, zero data retention across our stack for images. So even let’s say your kid take a picture of themselves, you cannot retrieve that picture even you want. You can ping me through my personal email. You cannot find that picture anymore. And OpenAI and Google signed a contract with us to burn a picture immediately, like zero data retention on all the usages. But overall, I do thinkGrace Shao (38:48)See.Reni Cao (38:51)It’s the company’s responsibility to introduce technologies to family and the family should hold a high bar there for sure. Because like the AI is so early and it’s way too powerful in certain way. And it’s like a kind of like a black box in certain way in a lot of different ways. that I definitely, I’m not a, I’m not that one of the technologies that wanted to like, you know, glorify AI and it is the future and stuff like that. comes with a lot of risk, especially like unproven.aspect how it impacts the children’s cognitive development and something like that. That’s also a reason why we work with researchers and professors ⁓ closely like in Mount Eucalon from UCSF and Harvard professors doing education and doing research using text. I do think there is a substantial risk here such that theAnd we as the entrepreneurs and we as the parents, we need to hold a high bar for ourselves and roll out things one by one. So I guess that’s why you will hear more about like, oh, that’s like, you could have done this. You could have made it more engaging. You will hear this much more often than you’d be like, oh, there is like an incident because, you know, we always prioritize, you know, safety first. We’d rather the device to be boring in certain way rather than introducing consequences that we don’t understand.So I think there’s a very interesting dynamic between the Western and Eastern in terms of their views about technology. And I don’t think it’s a family parenting only. It’s also even a whole society, general perceptions. Happy to chat about that, but maybe it’s a little bit off topic here. Yeah.Grace Shao (40:12)comes from the mindful design as well.⁓ No,we can definitely talk about that a little bit, but I kind of just follow up on what you just said. So how should a parent evaluate an AI device or tech device when they are purchasing for children, right? ⁓ I’m sure there are different devices out there, maybe not exactly doing the same thing as what you’re doing, but other devices are tech native ⁓ or AI enabled for children. How should parents kind of go about this?Reni Cao (40:52)I’m not a parenting coach. I will share my views. Number one, do think we should bias, we should start from our needs first. Maybe let me put it this way. Don’t get carried away with all the possibilities of the AI. Ask yourself, what is the unresolved parenting needs you have and find solution there. Rather than, this AIX, then that’s just to buy that AI device and give it a try. That’s number one, I would adopt that.Number two, I do think it’s important to see what a company’s method is. They definitely put their methodology somewhere, their belief somewhere, their principles somewhere, they’re kind of like, like how you ask me about how we use ALM. I believe that the parents should definitely hold the company accountable to explain those details and ask, verify, and that’s crucial step. That’s the due diligence on them, right? And I do think thatFinally, for any sort of AI product, I actually even think the parents doesn’t have to be getting into this searching and validating mental model. They could literally build their own in some sense. Given all the agent codings rising up and reducing the piece cost of software so low, I do think for lot of stuff, they should try.to accommodate their own parenting needs in certain ways. Like I saw tons of the parents go into cloud code generating like a, know, almost like a story writer for their daughter. That’s actually my previous colleague at Wish. And it was awesome. It’s just different blanks to fill in. It’s kind of mad lips type of like a story. I do think the parents can change also their way that they are in the autonomy right now to build whatever they want to build.Having said, it’s still a little bit of kind of like a Silicon Valley bubble type of answer, because honestly, in the world, the adoption of a cloud code is probably less than 2 % or 1%, I’m pretty sure. But I do think I would encourage parents to use AI themselves and explore a boundary, it can do, what it can does well, what it doesn’t. So then, the kind of I make a decision from there.Grace Shao (42:45)Yeah, no, I I appreciate that. It’s a very like thoughtful answer because it’s not just like A or B. of the day, think it’s parenting itself is so personal. It’s on how your family dynamics work, how you prioritize your time, how you want to parent. So when you want to buy technology for your children or incorporate that into their lives, it’s also a personal decision. I wanted to ask, actually, do you have any good case studies to share with us just a little bit?Reni Cao (43:11)We have quite a lot. What aspect, what, what type of case does he want?Grace Shao (43:14)Just like, I don’t know, like things that unexpected people use. For me, I mean, by default, just assume, yeah, people use it in urban areas, right? But then I think when I met you, you said, actually a lot of people use them, you know, in unexpected places, like orders come through all over.Reni Cao (43:19)Alright, I’ll give you one.One of, I immediately think about one thing, one, almost like ⁓ close to 5 % of our users, they bought Dex to help with speech delay. That’s something we never anticipated, but those parents are very frustrated with all the, as we call it, sometimes autism tech or the speech therapy tech there. It’s not meeting their bar and they saw Dex, they’d be like, I would try everything right now for my kid. And surprisingly Dex helped them.and it makes them real happy. And you can find actually all those real reviews in our review sections. Quite a few family mentioned that their kids refuse to speak certain languages or just even English, but that’s kind of necessitate the language as a fun activities. And all of a sudden, the kids start to open up and speak more, and the parents are really happy about it. This same exact story happened with my co-founder, who is really worried about his, at that time, two-year-old young son having speech delay.But I want to disclose the name, but the song he actually first time spoken like coherent like ⁓ Chinese phrases using Dex and he caught it on a video. That was one of the most wholesome moment of our kind of like a user feedback in our channel. And right now we’re actually ⁓ volunteering to develop this special need mode. That’s kind of like, you know, customizing to special needs children.especially like April is the world kind of like autism awareness month. And yeah, we just want to do it. And we want to donate to Dextre researchers and speech therapists to help us do it. This is a totally kind of like a side quest, but it just like give us it gives us so much kind of like energy. You’re thinking about technology can be used in a way that’s like immensely helpful.Grace Shao (45:00)That’s amazing.Yeah, and something unexpected, right? Okay, I think I want to wrap up our conversation because I don’t take up too much of your time, but I do want to ask you one big macro question. With you working on whether you like to call it physical AI or not, essentially like a physical product hardware time software, how do we understand that trend going forward? Do you think AI will be essentially integrated, plugged in to more more hardware devices? What’s your view on that?Reni Cao (45:33)I do think there is a consensus that every wave of software technology revolution, there will be kind of like a device revolution following that. We are at the tipping point there. That’s like people starts to reimagine, where is this? this cloud? Is this the recording card? Maybe it should be a separate, like an ⁓ AI. Or this is a sort of like a little pendant that can kind of like ultimately listen to your life, help you organizing. I do think we’re at the ⁓the dawn of a next wave of hardware. But it’s less about we’re doing the hardware because of the, I do think this is a software or technology driven type of hardware revolution out there. I do anticipate that. I do at least what I’m 100 % sure is like smartphones are not designed for children. Tablets are not designed for children. Families deserve something built with their interest.their needs in the center of the spotlight. And I see that happening. And that’s why we started this company. And I bet there’s going to be tons more use cases there.Grace Shao (46:33)No, amazing. Thank you. I think ⁓ one last thing. Is there anything I missed or anything you would like to share with us?Reni Cao (46:39)By the way, I time. If you want to turn through all the questions, I’m happy to be here. I don’t have anything else after this meeting.Grace Shao (46:44)no, don’t worry. think it’s a lot of times like I use them as prompts. But you know, when we’re chatting, like we actually covered most of it, you know. ⁓ Yeah, is there anything you think we missed? But from my end, like I feel like I covered most of it. You know, we did technology, we talked about children, AI philosophy, talk a bit about your business model.Reni Cao (46:50)Yeah. Yeah.Yeah.I do think you would want to talk about. Yeah, go ahead. Go with one last one, and I have one for you. Yes, go ahead. Ask yours first.Grace Shao (47:04)I think one last one. You go.So I wantto ask you one last question, which is a question I ask every guest that comes on the show. What is one differentiated view you hold? I feel like your whole thesis around devices right now on the market are not made for children is already a differentiated view. But is there anything else you think that you hold that’s non-consensus?Reni Cao (47:29)Yes, with this view, I got beaten up so many times, but I still got to say it, right? I believe that education should not be cookie cutter. It should be highly personalized. So is entertainment. So is the parenting software. And we’re about to enter the golden age. Finally, this is becoming the reality. And let me say it this way. You look at a school in the US, how you tell the school is good or not, you look at one ratio. It’s called a teacher-student ratio.One teacher taking care of less, but why? Because then the teacher can accommodate, individualize the needs. I actually have a very radical view in terms of our education system is definitely lagging, significantly lagging against how our society evolves, how the technology evolves. It’s still a one size fit all and industrial way.to handle education, handle like, you know, testing, standard testing. It hasn’t really changed in the past couple of decades, but the world is a different place now. And I guess my view is like, it shouldn’t be that. The default shouldn’t be that. The default is like every kid should almost have their personalized tutor and the playmate that deeply understand them. Unfortunately, that’s impossible before, resources-wise. But I guess we need to strive to get there.as a race, as a humanity. Because each kids just come up, come with their own spark.that will miss out the window to make that spark their lifelong journey. But I’m not trying to attack on educators or school systems, something like that. I just feel like there needs to be more forces from the society, especially from the tech side, to help together build this alternative, enhanced of like a system that really delivers individualized education.Sometimes I use the word scaled homeschooling. And you cannot imagine how much people hate that. people are like, homeschooling, you’re taking away the social aspect of it. People are very constrained on the vocabulary of how they describe things. But I guess when I say homeschooling, it’s not about keeping the kids at school and hiring a teacher. And that specific process right now, I’m talking about really meet children where they are in terms of their growth, in terms of their needs, in terms of the skills they’re going to develop.I call that a differentiator, but maybe actually lot of people will share the same views. I’ll be happy to know who shared the same view and please join us in the journey. Follow us along.Grace Shao (49:49)I definitelythink that view is definitely, feel anecdotally a lot more prevalent in SF when I visit. I’ve met other people like yourself, other people in the tech space or, you know, investors who are embracing this idea of modern homeschooling. And they say the same thing. They’re like, we don’t like to use the word homeschooling because, it sounds like a bit more cultish, but it really isn’t right. Like it’s really focusing on individual ⁓ growth.I think it’s amazing because I also think it’s because Silicon Valley itself kind of harbors this kind of growth and mentality and that the fact that people can succeed without degrees, people can succeed by building different things, people can succeed in just being different and being themselves, but the best version of themselves have always been, I think, what drives a lot of people who want to go to Silicon Valley because it’s like in many ways, as a mayor, talk to see like the best version of my talk to see, right? I thinkin East Asia, even as I put my kid in school right now, I find people definitely a lot less like that minded. ⁓ I don’t know if it’s a cultural thing because like, know, for you, you know, I grew up in Canada for me, I always felt like, you know, having that freedom to learn, explore when you’re young, which is more Western kind of way of, I guess, education was good. But I think a lot of peers here actually believe that, you know,for the first like say eight to 10 years, that foundational education should be drilled in. know, ⁓ grit should be taught, discipline should be taught. But it’s very interesting because it does kind of, I guess, manufacture different kinds of stereotypes. And I think it’s fascinating. And I think one more comment on that, I know this conversation has been more personal than we thought it would be, but I love it, you know.I don’t really get to talk about motherhood that much in my podcast. It’s usually about tech and bros and tech bros and about, ⁓ and about finance. but I think even, you know, when you have kids, people talk so much about nature versus nurture. And what I realized is I was shocked to see the nature come through, as young as like six to eight months in a child.Reni Cao (51:36)YouGrace Shao (51:53)their personality starts coming through and by the time they’re one to one and a half, they start kind of babbling, start demanding things. I realize 80 % of it is all nature. It’s like their preferences for how they socialize, their preferences of even noise, even you can realize like your point, your taste. You’re going to find a six months old who just wants to sit in a corner in a play group who just wants to flip through books literally and just undisturbed. You’re going to find someone who’s screaming in the middle of the whole group.you’re gonna find my daughter who’s rolling over everyone and just like trying to knock everyone out. And I don’t know why. You know, you’re gonna realize all of it is nature. And even I believe agency, autonomy, grit, and desire to actually succeed, that itself is nature. And I don’t think you’d be taught. And I think this is a bit controversial. But definitely I think my husband and I have been thinking a lot about this. We’re like, we can just provide them what we can. But there is no...point of even pushing them when they don’t want certain things. the best is to push them in a direction that they want to be pushed and they will tell you. I think this is like kind of the difference in our generation of parents. yeah. Reni, thank you so much. ⁓ Yeah, go on.Reni Cao (52:49)Exactly.Yeah, but can I, I know thisis over time, but can I add one last comment towards what you say? But I think what you said, especially growing up in East Asian, like, you know, education system, it has been industrial for a good reason, right? At a time where stuff like AI doesn’t exist, the most effective way,Grace Shao (53:03)No, of course, of course.Reni Cao (53:19)to develop fundamental knowledge workers, plus finishing the job of dividing the children into different segments and give them different levels of education. That education system works perfectly. I entrance exam, as I’m talking about, taking standard tests and stuff like that. But all we know that is AI is sweeping through all the knowledge works.and specialized in knowledge works, honestly, Asian parents like favorite jobs, like being a doctor, especially radiologist, you know, and or being a lawyer, you’ve got to start somewhere as associate. Now it’s getting kind of like his hardest. The world has already changed. The tsunami already hits. But I don’t think people actually understand the level of the s**t. A lot of like everyday people in the world, they haven’t felt.this like a tsunami, right? So when you say you want to kind of like, you know, like find define your children’s nature and push them towards kind of like what they are intrinsically motivated about and give them resources to set them up for success, building grades are on the way. I do believe that I think I will 100 % agree with you that it will become the most fundamental aspect or element of education in the next like five years or even sooner to be fair.That’s why I don’t send my... to put it in a simple term. I don’t send my daughter to Kumon. I don’t want my daughter to do Russian math. I never benchmark her against like, oh, like the other kids can read at the age of like three and a half. Why don’t you? Actually, I don’t because I fully understand that kids have their own time zone. Kids have their own spark. All you need to do is think deeply to define that, to understand that, understand why my daughter sometimes is super sensitive, understand why sometimes she got frustrated and want to hit.Grace Shao (54:31)I feel very validated.Reni Cao (55:00)Don’t take that on a surface level with the other tools you have. Go deep, understand that, and build these programs that’s personalized to her and help her. And I think like this is why I if I, talk about the word of Nei Juan a lot. If I have to dream on anything, right? I have to like a rather like ruthless compete on anything. I complete the deaf understanding of my daughter rather than anything else.Grace Shao (55:03)100%.Reni Cao (55:26)Because I actually think that’s the thing people gloss over. People must be like, education is just checklist. You got to check, check, check, check. And there is a better checkbox. Like Ivy League school, there’s a OK checkbox. There’s a worse checkbox. Forget about a checklist. That checklist is obsolete already. So I respect. I think we vibe together in terms of our schools of parenting.Grace Shao (55:33)Yeah.Yeah, 100%. No, I agree with you.parenting style. Yeah, yeah,yeah.Reni Cao (55:51)But you’re so fully intuitive. don’t know whether I’m right or wrong, but this is what I firmly believe in. And I believe someone’s going to join this journey.Grace Shao (55:58)I think there’s more people who are aware, especially people who are more plugged in with the technology because they realize how fundamental society will change. I just thought about when we were young, I’m sure your parents also told you to go to university, go to this, go to that, right? For sure there was a hierarchy in their mind, what kind of school you should go to, what kind of degree you should get. Now I really don’t think that’s the case. Actually, a lot of my readers would even know.my dad really forced me, well, pushed me, encouraged me to go into finance. And at one point he was like, if you don’t study finance and don’t work in finance, you’re not like following my footsteps and blah, blah, blah, blah. Right. And it was a very, it became a personal reason to do it. It’s not because I wanted to, or I was good at it. And there was actually a battle between us being like, I want to go into journalism. And he’s like, no, I was like, no, I’m going to go to journalism. He’s like, I’m not going to pay for it. You figure it out. But the beauty of it is actually found a way resourceful enough to get a full time, full scholarship.And I still want to journalism. Again, I recognize how lucky I was. I I found the opportunity to do that. But most kids actually just end up then doing what their parents told them to do and they never, and they never actually live their best life or become the best versions of themselves because they’re doing something not actually fundamental.Reni Cao (57:08)you hit a very critical, I think it’s a background or context. There wasn’t an abundance before, right? Growing up, let’s say in the eighties, It’s a relatively kind of like a society. It’s relatively kind of like not that sort of like, you you wouldn’t call it abundance at the time. Let me just put it that way, right? You still need to compete for stability, compete for resources. That’s why there’s a rat race in education, which I totally understand. That’s kind of like.It’s like a whole economy there, right? But I think that changed. No matter what we’re talking about, like, I mean, in China, I’m talking about US right now, I think abundance really will hit at some point of time. At that time, the challenge shifted from how can I avoid getting into a property or like ⁓ job loss towards kind of like, how can I find the meaning of my life? And how do I deal with this kind of like a journey?It’s a generational theme there. It’s very funny that your dad wants you to go into finance rather than journalism. mean, for those who understand Chinese internet a little bit recently, there has been a famous influencer called Zhang Xuefeng. He almost helps everyone to pick their college major. And one of the college major he hates the most and advice everyone to not go to is actually journalism at the end of the day. Because it’s just not a...stereotypically stable job that can make a lot of money, that can give you social status, quote unquote stuff like that. But I think that’s a lack of view of things. Our dad doesn’t know how our skill landscape is to be 20 years later, just like we’re not going to know what our kid is going to deal with. So we need to give them a more sort of generalize the resources and the skills at grit to survive. And it’s funny enough.I made my, I named my daughter Simone because we are big fans of Simone de Beaufort, the kind of like, you know, foundational philosopher of feminism and a lot of like a sociology thoughts. So, you know, if I have, if I have to give a set of expectation for my daughter, I want her to actually do something that’s not commonly seen as a very, like a prosperous or stable kind of like career ladder. I want her to do something that’s kind of likehas a mission and some sort of like outlier type of journey. let’s see how that goes. She’s so young, so who knows.Grace Shao (59:20)I love it. All right. Thank you so much, Reni.Reni Cao (59:23)Thank you. AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  There's more to Korea than just chips. TheVentures CIO on the country's AI stack 11.05.2026 54λ

  In this episode, I spoke to a leading South Korea VC, TheVentures’ CIO Ethan Cho. He argues that South Korea’s low fertility rate and aging population put pressure on Korea to be one of the world’s fastest adopters of AI technology, similar to its rapid embrace of high-speed internet in the early 2000s. While not a leader in foundational LLMs like the US or China, Korea’s strength lies in application and adaptation, particularly in B2C areas like personalized agents and commerce, where cultural familiarity with chatbots and digital transactions lowers resistance.The Korean startup capital funding landscape is shaped by three forces: Chaebols (Samsung, SK, Hyundai), the government, and VC firms. CVCs from Chaebols tend to reinforce existing semiconductor and hardware value chains rather than explore tangential innovation. To counter this, the Korean government has become a dominant LP through initiatives like “Everybody’s Entrepreneurship,” injecting capital to encourage novice founders. On sovereign AI, he believes the government’s push is less about global dominance and more about securing sensitive areas like finance and defense, though he warns that domestically-built software has historically struggled to scale beyond Korea.Ethan is shifting focus from purely domestic champions to founders with global ambition but local execution, often Koreans educated abroad who return dissatisfied with traditional jobs. He wants to back ventures that change the world, not just build another food delivery app. He also recognizes key opportunity areas, including defense tech, K-beauty, fashion, and mental health, as society adopts AI at scale.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently. Season two will host a series of guests from early-stage investing, as well as builders, founders, and product managers.For more information on the podcast series, see here.To find the previous episodes of Differentiated Understanding, see here.Chapters00:00 Introduction to Ethan Cho and His Journey02:47 Korea’s Role in the Global AI Supply Chain05:24 Cultural Attitudes Towards AI in South Korea11:06 Government Initiatives and Sovereign AI16:37 The Future of Commerce and AI Integration28:03 Consumer Behavior and AI Adoption28:43 Enterprise AI Solutions in Banking and Manufacturing33:39 Investing in Founders: The New Generation of Entrepreneurs39:39 Korea’s Future Exports: AI and Beyond41:41 K-Beauty and K-Fashion: Cultural Exports45:15 The Future of Mental Health in the AI Era49:53 The Limitations of AI and Human ExperienceAI- generated Transcript Grace Shao (00:00)As mentioned, our guest today is Ethan Cho. He has been active in the Korean VC space for over a decade with experience in the venture investing arms at Qualcomm, Google, Samsung and more. Now as a partner at the ventures, he leads a team focused on finding and nurturing the next generation of AI native startups. Ethan, thank you so much for joining us. So good to have you.Ethan Cho (00:18)Thank you, Grace. I’m very excited to be on the show and I would love to discuss with you more in detail.Grace Shao (00:24)Yeah, to start with, tell us about yourself. Tell us about venture investing in South Korea and the firm’s background.Ethan Cho (00:31)Sure. So I was born in Korea. I kind of moved internationally quite a bit. I moved to England when I was kid, when I was four years old. That was where I first learned my English, lived in England about four years, came back to Korea, then moved to Hungary, lived in Budapest for a year, came back to Korea again, spent the next 20 years in Korea, moved to the States, lived in New York for ⁓ my business school years and worked there for another year, came back to Korea after then. So I’ve been in and out of the country quite a bit.I loved startup investment very early in my career, so I wanted to move towards startup investment. I actually started out as a hedge fund analyst right after business school, but I quickly found out that I’m more interested in finding good companies and good stocks. So then I moved towards the private side, started with Samsung, and moved to Qualcomm, et cetera, et cetera.What fascinates me about Korean startups and startups in general is that everybody’s trying to change the world. I’m just such an honor to be a part of that and talking to entrepreneurs on a daily basis really excites me.Grace Shao (01:34)Awesome, I think it’s really interesting because you’ll definitely bring a very international perspective and not only just the Korean perspective and also kind of understand, you know, where a lot of our listeners are coming from as well. You have, you know, you have exposure to UK exposure to Europe, exposure to US. I think I want to ask you quickly, because you’ve actually worked in the public sector as in public investing, it’s kind of interesting because right now, obviously, the frenzy and the, you know, the global interest right now.Ethan Cho (01:57)Mm.Grace Shao (01:59)is in a lot of the big semi providers in South Korea, you know, focused on infrastructure layer that are public listed. At a high level, how should we think about the Korea’s role in the global AI supply chain? And then of course, we’ll shift gears into talking about the startup scene that you’re passionate about.Ethan Cho (02:02)Yep. I think that’s a great question. think one of the very obvious factors of AI is memory because you can only use AI based on whatyou or the agent knows about you or any company. So because of that, the demand for memory is exponentially increasing. I think that’s definitely a blessing for the career semi-players and also the current industry as a whole. But at the same time, think nature has always found a way to become more efficient. So the capacity or the demand constraint will not beexisting forever. There’s going to be significant improvements such as Moore’s Law. There’s always an innovative solution that comes out every other year. So I think there’s definitely going to be a lot of exciting opportunities down the road, but it always evolves. So it’s going to be very different next year. It’s not going to be HBMs anymore. I think it’s going to be something else going down the road. We’ll have to see, but I think the trend is here, but the trend itself will also keep evolving down the road.Grace Shao (03:15)Awesome. So, you know, if we have a really candid assessment of what’s happening in Korea right now, what’s genuinely really strong do you think? Like the chips are strong. ⁓ What do you think that weaker maybe compared to, you know, China and the US, you know, from outsider lens, is it really the LLM labs kind of or is it diffusion? What’s happening?Ethan Cho (03:24)Hmm.Yeah, I mean, it’s a complex situation. I think one thing that a lot of people quote and one kind of fate that we cannot deny is that we have a very low fertility rate. So the birth rate is decreasing very fast. We have a very rapidly aging population. A lot of people think of this just as a curse. I think there is an aspect to it that it’s a blessing in disguise because we are one of the countries that most desperately needs AI and robotics.And because of that, I think we will be one of the more adapting or welcoming countries for AI and robotics. If we go back to the early 2000s, we were one of the countries that adapted most rapidly to high speed internet as well as mobile technology.because we had to, we talk a lot on the phone, obviously. So because of that kind of demographic instinct, we were one of the much faster adapters to that technology. I think that’s gonna repeat in AI and robotics. As you mentioned, I think that AI and robotics is definitely not, we’re not the strongest when it comes to AI robotics in the world.But in terms of adapting and using it for actual use cases, we may be one of the very strong countries. So I think there’s a lot of challenges and opportunities ahead of us.Grace Shao (04:50)That’s really interesting. think you hit something that like, you know, people are starting to pick up in the West, which is like in East Asia in general, the embrace of technology is a lot more optimistic. Some come from very realistic reasons. Like you mentioned, whether it’s in China or Japan or Korea, there is a potential labor shortage that’s coming to the next generation, right? But not only so, I think just in terms of culture and social sentiment also feels that way. So if you have to give like a high level kind of assessment onYou know, just the cultural attitude and political attitude towards AI, what does it feel like on the ground in South Korea?Ethan Cho (05:24)I think AI itself, I think people think of AI in different forms, obviously. I think as far as I know, China thinks of AI closer to robotics. The US thinks of it as, I don’t know, maybe a chatbot or something that they use for the industrial usage. In Korea, as far as I’ve experienced, I think it’s more about becoming a personalized kind of agent, not agent.per se in terms of doing purchasing or actual daily tasks. We always had kind of chat bots, especially for instance, for our financial system or the banking system, we’ve always used CS bots very frequently. So we’re very used to it. So I guess there’s less resistance when it comes to adopting or adapting bots or AI featured functionality, especially in the B2C area. So although...When we say AI just by the name, it could sound creepy. I think it’s already very well embedded in the Korean startup ecosystem and the overall society as well.Grace Shao (06:27)That’s interesting. So you did kind of touch on one thing, you know, the Chinese view this way, the Americans view that way, you know, definitely there’s a bit of a difference in terms of how maybe AI is being seen as whether it’s a political agenda or economic ⁓ aggregator or, you know, how it’s being diffused to the seaside. So in that sense, ⁓ my question is, is Korea building domestic AI champions or is it more right now kind of working around, you know, working on top of U.S. frontier models?or leveraging a lot of the Chinese open source models. Like how do we understand that in the ecosystem?Ethan Cho (07:01)Honestly, this is a personal kind of statement and my personal observation. I think it’s all of the above. I think we were going to talk about this anyways, but the sovereign models is something that the Korean government really wants. I kind of understand it in a way. The SOTA models, the state of the art models are obviously the ones that the enterprises want to use.But at the same time, think a lot more people and developers are looking into Chinese open source models, especially with the recent changes in cloud code and Gemini and everybody, who are basically hiking their token prices. It’s getting more and more expensive to actually do recurring work. And at the same time, think, including myself, a lot of developers are quickly finding out that the fine tuning of the models can only be achieved by very,⁓ almost redundant loop work, which can only be achieved through open sources if it is to meet economic sense. So right now, for instance, if we use a certain American LLM model to do these hundreds and thousands of ⁓ very repeated work, that costs a lot. So from an individual standpoint, that’s not really easy to achieve. So I think everybody’s trying to find that sweet spot of mixing those three models.Grace Shao (08:16)That makes a lot of sense. I think for stars, especially the ones that you work with, know, the economic driver is probably one of the biggest reasons why they choose what. So I do want to save the sovereign AI kind of piece for a bit later to help our listeners understand, you know, the create ecosystem a bit better. We all hear about Chibbles. We hear about, you know, obviously the big tech like the Samsuns and the whatnots in Eskihainix right now that are getting a lot of attention, right? Help us just even understand how these different companies and the startups, how they work together. Because for example, in China, a lot of that big tech are actually the incubators and initial investors of even the startups. So even the leading LLM labs, they actually have taken 10 cent Alibaba money. In the US, it does feel a bit different. There’s vast amount of venture capital money that are kind of funding the current growth rate of OpenAI and anthropics of the world. How does it?Ethan Cho (08:54)Mm-hmm.Grace Shao (09:09)ecosystem work in South Korea.Ethan Cho (09:11)So I think there was ⁓ quite a few phases of evolution. when the startups were really founded, that was, I don’t know, that was like late 90s. Those were very purely internet domains, internet online communities. Like that had not a lot to do with the Chebals. But then came mobile technology and everybody was starting to invent something on mobile. That quickly got the...Interest from the big Chebos, but actually as far as I know there were some Interests in very early on in neighbor and cacao by all these like really large companies in Korea But they never actually fully understood what that what that was and they kind of let them grow Which was a blessing for us at the end of the day. So neighbor and cacao was you know established and they grew like crazy after that thethe big companies quickly found out that, we need that DNA of innovation. They started to set up their own VC firms. They started to set up their own accelerators and everything. But as a typical CVC does, they inject a lot of money into their interest area, but not so much in let’s say, tangential areas. So because of that, there are definitely a strong value chain around semiconductors, for instance. But that kind of reinforces thealready existing ecosystem of the chaebols, which is not exactly what the startups are intended to do. So there’s kind of pros and cons there. And on top of that, after the capital was kind of concentrated into that value chain, the government kind of now is more active in kind of leading investments. large chunk of investments in Korea is led by the government.by the mother fund or fund of funds of Korea injecting money into the ecosystem and the other funds matching to that. So there’s a layer of chaebols, there’s the government and also the capital firms who are also ⁓ acting as LPs for the Korean startup ecosystem.Grace Shao (11:06)Yeah, that’s a perfect segue into understanding, you know, the government’s play. So I visited Korea just recently, I think last November, and, you know, it seems like there’s a huge policy push as well in incorporating AI into the everyday everything. And, you know, it’s top down driven. And like you said, there’s capital also injection. So how do we understand the government’s current priorities in terms of embracing AI? How do we understand sovereign AI andwhat kind of role it plays in the economy South Korea going forward.Ethan Cho (11:38)⁓ I think the government is definitely ⁓ making a very interesting and important bet. So there’s this huge initiative called Everybody’s Entrepreneurship, loosely translated into English. The government is actually injecting a lot of money into the ecosystem by giving money to...people who want to become entrepreneurs, who are novice entrepreneurs, first time entrepreneurs. I think it’s a good thing that a lot of people are trying out their ideas at the end of the day. The downside honestly is that entrepreneurship isn’t for everybody. So there’s gonna be people who learn their lessons the hard way, but still I think all in all, it’s gonna be a positive impact on the overall ecosystem. I think...⁓ The AI drive is definitely very serious for the government. As we mentioned earlier, there’s a labor shortage coming in. I think ⁓ East Asia most of the time has a little bit of issue with immigrants. We don’t shy away from immigration, but I think traditionally we don’t have the most welcoming immigration system compared to the States, for instance. So we’re trying to buy some time there, I think.And I think because the strongest point of the industry, as we also mentioned earlier, is semiconductor and hardware and technology, we want to build upon that. And because of that, I think AI seems to be a very interesting and promising area for the government and Korea as a country overall.Grace Shao (13:03)How do we understand sovereign AI though? Like what is the, I guess, reason for like, you know, maybe the non too large company, sorry, too large economies to really start pursuing this? Because we’re seeing this kind of rhetoric in the Middle East as well. You know, a lot of the local governments are really pushing sovereign AI. South Korea for sure has been openly talking about this. think, you know, a lot of European nations are also thinking about this. Is this just for, I guess, owning?Ethan Cho (13:06)Hmm.Grace Shao (13:30)the future infrastructure or how do we understand this?Ethan Cho (13:33)I think that’s one point. I think owning the future infrastructure is one. But I think if people are realistic, think we don’t want the world, we don’t hope the world is going to use our own sovereign AI. I don’t think that’s the case. What I’m expecting or I believe that the government people are wanting is that to use sovereign AI in very sensitive areas, such as our financial backbone, for instance, orYou know, because Korea is technically set or on the defense part, maybe we’ll use that for that purpose specifically. I think in the early days when everybody in Korea started to talk about sovereign AI, I was actually less persuaded. The problem is, or the status is, as we see all these leaks all over the place, like even for the top, you know, bleeding edge,builders like Anthropic or OpenAI, there’s always issues here and there. And it kind of shows. I’m not saying that sovereign AI is going to be perfect either. They’re going to have issues too. But if a foreigner comes, a foreign entity comes in and kind of screws up an operation, that kind of blame and whatwhat something domestic spills over. There’s going to be a different kind of anxiety in the society, I guess. So that’s maybe the angle that they’re kind of anticipating. But I mean, I am worried a little bit too, because I’ve seen multiple cases of software built in Korea domestically, which has never been successful to Korea. And it just has been kind of a wanted wonder just in Korea. I just hope that doesn’t repeat. But we’ll have to see.Grace Shao (15:06)Actually, this is not completely rated to AI, but just on that note, why do you think a lot of the times like Korean companies are huge? Like you just mentioned Naver and like Kakao or like, you know, Japan, and LINE and China that we chat when not like these internet companies never really go abroad. That’s just intellectual curious question just on the topic.Ethan Cho (15:16)Mm.I’m a kind of linguistics buff. So I actually think the reason is in the language. The language that we speak is just different from English. because of that, think the like Naver, Kakao, WeChat and Line are all basically rooted in language. And because of that, that’s just universally different from WhatsApp, for instance. Like it’s not language per se, but if you look at WhatsApp, how they control their UI UX, for me at least, is very boring.Grace Shao (15:27)Mm.Ethan Cho (15:52)⁓ I would prefer a cacao or lime or WeChat much over WhatsApp if I could choose without being specific in which geography. So I think there’s a cultural preference, a very strong cultural preference that really is hard to translate across territories.Grace Shao (16:09)Okay, that’s an interesting take. Yeah, because I think it’s interesting because like, it’s basically the West has this one or even like, you know, Africa based Southeast Asia, they all fall under the American big tech kind of umbrella. And then Korea, Japan and China’s have such strong domestic players. Actually, on the note on you know, the consumer side of things, what are some interesting trends you’re seeing out of South Korea in terms of consumer AI right now? What are some companies you’re investing in that are in the consumer AI space?Ethan Cho (16:16)Yeah.Consumer AI, think, is still at a very early stage of growing. I think right now the most used cases that I see with my bare eyes are actually foreign tourists coming to Korea, visiting like big K-Beauty.department stores like Olive Young, and they go and get their skin scanned and they analyze it with AI and give recommendations to basically ads. But still, I think that’s a very clever way to scientifically analyze the customer demand. I think a lot of players, and I see a lot of players trying to replicate that into basically recommendation engines. Personally, I think that’sclever, but it’s not good enough. It has to get better. The trade-off there is obviously privacy. So if you want a Uber personalized recommendation, you have to somehow yield on the privacy part. think we’re still not clear on where is the kind of safety line. So I think they’re still kind of struggling towards that. We have invested mostly these days in consumer brands because how⁓ I think of it at least, is regardless of which AI becomes the winner or winners, I think as long as we have the best product in our portfolio, if the AI is clever enough, it will choose that product. So before actually deciding which AI algorithm will actually win the war, think we’re trying to get hold of the monuments before the, know, who,before we decide who becomes the winner of the war altogether.Grace Shao (18:12)So you’re looking at brands as in like, like retail brands. What are you looking at? Like, okay.Ethan Cho (18:16)Yes, yes, for now,yes. But at the same time on the kind of the hardcore AI part, we also have invested in sovereign AI companies like Trillium Labs, which actually develops SLMs instead of LLMs. We’ve also invested in a drone company called Bone AI, which does physical AI using drones and they’re targeting the Korean big defense industry. So that’s kind of themore of the hardcore AI part that we’re looking into. We’re still waiting for that sweet spot where consumer meets AI. I think that’s still kind of in their very early stage in Korea.Grace Shao (18:46)I see. I just want to say it’s so funny you used the Olive Young example, it’s really topical. So this is really a bit of a rant, but my friends and I were saying we need to go to South Korea to do the color palette, right? And all, you know, all the girls are like raging about this right now. I literally asked Claude Coe today to do it for me and it gave me the whole like color palette assessment. I was like, wow, I just saved myself a flight and a trip to South Korea. So definitely can see like there are consumer uses in that end, but I guess what is the monetization from that bud, right?Grace Shao (19:22)So that I can see how that will be hard to invest in that space. On the consumer end, know, again, I read headlines about South Korea, right? And, you know, I hear about companion bots being really big. I know actually even in China, they were group bots. They were like so-called boyfriend, girlfriend bots. And to your point, you know, South Korea, China, like even like a lot of East Asian countries are in are all faced with this issue where there’s mass urbanization.Grace Shao (19:48)loneliness issue everyone is like, you know faced with evolution and competition so they don’t have human companion and Do you see this as a trend and do you see this as something that potentially would not be actually within the cacau’s and the lines of the world that could be a spun-off on its own and to fall off of that I was just even just kind of thinking because Korea has so much IP right now in obviously k-pop and k-dramaEthan Cho (19:57)Hmm.Grace Shao (20:16)would that potentially be a vertical where they can tap into basically creating, you know, like companion bots, based on existing celebrities.Ethan Cho (20:25)I mean, I think yes on both is the short answer. I think the boyfriend bots, girlfriend bots are very popular in Korea. I think my son is also using one. I haven’t talked about that openly, but I think so. And yes, I think that Kakao and Neighbor would be very cautious about adopting that technology into their existing platform. There can be some...many opportunities to abuse that. People, as you know, once these bots are online, the first thing that everybody tries to do is abuse it one way or another. So I think a cow and neighbor would probably shy away from that and go into commerce, which is always what they wanted. We’re seeing more specialized startups that are just doing this boyfriend, girlfriend bots.like some are more adult focused, some are more teenager focused. So there’s definitely kind of a breed that’s coming out of that. On the K-pop and K-drama bots, I think that’s something that a lot of companies have worked on for quite a while. For instance, like Weverse, is the, it’s the entertainment company that basically ⁓ relates to all K-pop stars. They have their own like personas.So they actually provide not just only conversations on bots. I think they also give artificial voice calls. They’re already there. So you can have a conversation with your favorite star. It’s just not realistic enough yet. But I think they’re getting there. So ⁓ that’s definitely already happening. I think on the IP side, personally, think it’s more about how can you make these into really long-lasting legacies? Even for BTS and like...girls generation, which are the you know, the idols of the day. I think they’ve only been around for 10, 20 years. Like, can we make this into like decades long, right? Like a legend, like can we actually make this into something that goes through generations, not just decades? That’s a big homework for us to figure out and make them kind of timeless.Grace Shao (22:24)No, I totally see that. It feels like a Black Mirror episode with the Miley Cyrus ⁓ kind of fake doll as well. But I think to your point, there’s the IP issue, there’s obviously the security issue. There’s the psychosis issue. This is like a much bigger issue. I think that we require regulators to work with businesses, right? I want to kind of move our lens to the enterprise side of things. You you mentioned just now like Naver and Kakao wereGrace Shao (22:50)looking into maybe going to agentic AI and maybe even to commerce. Are we looking at something like what Alibaba is trying to do where you have agenda commerce through a one entry point, you you interface with a chat bot next thing you know, like a bubble teas at your door. Like, is that the future of commerce you think or what are we talking about here?Ethan Cho (23:11)I think, I think, neighbor and Kakao are both in fierce competition with coupon coupon is the dominant e-commerce player in Korea. ⁓ I think it’s a real headache for both of them because coupon was kind of non-existent. They didn’t have a lot of user interface and neighbor and Kakao were kind of self satisfied that they dominated the user interface. But, here comes coupon and they just basically just crushed every.aspect of e-commerce and is by far the number one player in Korea. So that’s something that Naver and Kakao are trying to battle. The only difference, as far as I can see, that they can make is real-time purchases. If you want milk at your home the next day, coupang is much easier and much better. That’s kind of a fact. But for Naver and Kakao, because they have basically 24-7 access to your daily life, if they can actually monetize on that, I thinkThat’s their way to go. The competition there is also not non-existent. That’s a problem. YouTube’s there. TikTok’s coming along in Korea. TikTok’s still small in Korea, but it’s growing rapidly. that space is also... I think people think... Koreans are just so big YouTube fans. The YouTube dominance is so... They used to. Now they’re getting more used to short forms now.Grace Shao (24:15)Why is it small? Curious. Why is it small?So they like long form.Ethan Cho (24:30)And the TikTok trend is definitely coming along. But I think most YouTubers, so-called YouTubers in Korea, are long-form originated. So the trend is changing now. So yeah, it’s a little bit slower to adopt. yeah, sure. Oh, yeah.Grace Shao (24:40)And I wanted some context. So coupon is like an Amazon or like it sounds like a DoorDashand like how do we understand this just for our American audience or Western audience?Ethan Cho (24:51)Yeah,Coupang is, they actually literally say that they want to be the Amazon of Korea. So they are the Amazon equivalent in Korea. Their main business is e-commerce. They have Coupang Eats, which is DoorDash or Uber Eats. They have Coupang Play, which is the Amazon Prime. So they’re basically, Coupang people would hate me saying this, but it’s kind of like the Amazon replica in Korea.But they’re doing a fascinating job. Their killer feature is next day dawn delivery. So if you deliver, if you place your order by midnight, they’ll get the item to your doorstep before 5 a.m. So it’s marvelous for. It’s not just groceries. Yeah, so they’re very good at demand expectations. So they have a lot of warehouses in Korea. So they fulfill them in advance so that they can basically distribute almost within like four or fiveGrace Shao (25:29)It’s not just groceries, it could be anything.Ethan Cho (25:43)hours window, which is also honestly possible because Korea is not so big as a country.Grace Shao (25:49)But it sounds kind of like almost a JD.com business model as well. It’s like, but more high, high, more expensive. Excuse me.Ethan Cho (25:52)True. Yeah, yeah, I think I think that’s a fair. Yeah, that’s a fair comparison. Yes.Grace Shao (25:57)It’s a bit more expensive, right?Yeah, so I think looking at that, then, you know, there’s also this rumor, or I guess it’s actually been verified that South Korea was, fact, OpenAI’s largest enterprise market outside of the US, which is crazy, because like you just said, South Korea is not exactly that big of a country. Why is that? Who are the people buying up all these tokens?Ethan Cho (26:12)Hmm. I don’t have exact numbers, but when I first heard that I wasn’t too surprised because if you look at like Koreans are very used to buying tokens online. Like that’s why Korean gaming has beenor what used to be so big, especially in the mobile era, because people were just fine with buying items online, like purchasing it like crazy, which was kind of now it’s kind of standard, but like back in the days, like early in 2000s, like it was a very weird phenomenon if you look at it from a global standard. So because of that, think people are really, really fine with just buying tokens and buying memberships, which costs 100, $200. I think that’s kind of whatthe base layer, so the willingness to pay the first layer. The second layer is that Koreans love to build things. Just trying to build things so much with their own hands is one tendency that we strongly have. So because of that, I think most of that building tokens go to my stock portfolio optimizer or kind of tools like that for personal use. But I think people just like to try out a lot of things that kind of led to that consumption.Grace Shao (27:26)Mm-hmm.Ethan Cho (27:33)On the B2B side, think as far as I know, because the head of OpenAI Korea used to be one of my kind of bosses at Google Korea, he was, well, OpenAI was very aggressive making contracts with Samsung and SK very early on. So I don’t know how many tokens they’re using, but just thinking about how many employees they have, if they struck a good deal on a B2B,business, I think that would be a very significant portion of tokens being burned in Korea as well.Grace Shao (28:03)That’s pretty crazy. So basically you’re saying the the enterprise side, people have pretty strong connections and reasons to buy and the consumer side are just willing to shell out subscription dollars. That’s very different from, would say, the Chinese market where there’s just like not a lot of willingness to pay from consumers. And hence why we saw all the consumer apps in China were all free. So I actually want to ask on enterprise end. So what are we seeing people spend money on in terms of AI thatEthan Cho (28:14)Yep.Grace Shao (28:30)What are people trying to build? Are we looking at like also like on the enterprise and are they trying to solve co-pilot like solutions? Are they trying to serve customer service issues, manufacturing optimization? Like what are people really focused on?Ethan Cho (28:43)So just based on my experience with the companies, I think one thing is the banks are very serious about building the CS layer via AI. So they want to substitute a lot of that labor force into AI. I’m not sure whether that’s like how fast that can be optimized just because people are very demanding in Korea. know, when even if you use the traditional kind of phone CS, peopleend up basically talking to people. They demand to talk to an actual person instead of going through the automated call. So we’ll have to see how the ROI comes out on that part. I know that there’s a lot of AI being used for the semiconductor processing and producing process, but that’s just not public information. So we really don’t know how much is being used there. So that’s on the enterprise side. On the consumer side, think because of everybody’s⁓ entrepreneurship program that I mentioned earlier. I think there’s a lot of people that are trying to use a lot of cloud code, for instance, or codecs from OpenAI to build programs. There’s a lot of events actually held in Korea. Maybe every week there’s an event from OpenAI or Anthropic basically, which is like the cloud ambassadors night or the OpenAI something, something night. people, lot of... ⁓the AI builders are actually encouraging Korean builders to use their own tools by giving out a lot of free tokens actually, like thousands of dollars are given out as tokens just to nudge them into building. So there’s gonna be a lot more activity in that space for sure in Korea. And hopefully there’s gonna be something that’s really interesting coming out from that.Grace Shao (30:25)That’s interesting. I did want to ask, you kind of mentioned this earlier that you guys even invested in a drone company. South Korea obviously has a very strong manufacturing sector, home appliances, phones, cars. How are we seeing this whole, we have generalized this whole sector kind of lean into AI? Are we seeing physical AI being prioritized? Are we going to see? more robotics coming out of South Korea. How do I understand that?Ethan Cho (30:55)I think there’s still some uncertainty there because of the all of the among all the Korean robotics companies, I think the most technologically advanced one is Boston Dynamics, but that’s not a Korean Korean company, to be honest, right? Because it used to be an American company acquired by Hyundai Motor Company. So there’s that. There are quite a few robotic startups that are starting in Korea.Just because we have Samsung, Hynix, and Hyundai, think the manufacturing industry obviously is a great application area or a market to sell to. So we are seeing a lot of robotics company coming out from the university as well as startups. The big question here is will they scale? That’s kind of the pressing question. I mean, I think the companies, for instance, for Coupang,which does all the logistics. They’re heavily using robotics just as Amazon does. So those robotics are already deployed or are being deployed. But for instance, humanoid robots, which China is leading the way, I think that’s still a long way to go for us. And we’re trying to figure out what the application should be. So one interesting example, I think China has this too, but...We have all these little, really cute delivery robots going down the road and trying to get food to their neighbors. That’s an experiment that a lot of companies are running right now. We also have small police robots that are also running around just to do surveillance. I think it’s a cute initiative, but can this scale is going to be a big question for a lot of us. So I think this is also intertwined withautonomous driving landscape in Korea, which is still kind of not there yet. So I think there’s going to be a lot more of this going forward.Grace Shao (32:44)Yeah, no, actually on that note, I just was in Shenzhen last week and I saw one of these like, you know, street sweeping robots per se, stuck in a puddle. And it’s like, to your point, like they look cute or like, you know, you have little robots delivering your phone charger in hotels, but they’re not actually that scalable. And I don’t actually know if they’re that cost effective is the issue, right? Because, you know, sometimes hiring a person to sweep the floor, frankly, inEthan Cho (32:48)Hmm.Grace Shao (33:12)a market like China is actually not that costly compared to even deploying a robot like that and then having to, you know, maintain it. So I see your point. Okay, I think, you know, I want to shift our focus back to, you know, your bread and butter. And I really appreciate you patiently breaking down the ecosystem for me as an outsider who don’t understand South Korea that well. But as a venture capital investor right now in South Korea, what are your, I guess, most interested areas?What kind of founders do you really want to invest in? And are you looking at the founders more or the companies more? Let’s start with that.Ethan Cho (33:45)I am looking for founders. I’m looking for founders because I think there’s been a evolution of generation or a change of generation that I’m seeing. I see a lot of Korean.like in their 20s or their 30s who are educated abroad, come back to Korea and start working in Korea, not too happy about their job and trying to figure out what to do next. I just want those people to actually start something new and I want to kind of back them. I call that like global ambition, but local execution. I think that’s something that we need more.Until now, as you know, all the companies that we’ve mentioned throughout this conversation, like Naver, Kakao, Coupang, they’re all basically really focused on the Korean market, which was, you it’s good. But still, as we all know, Korea is not the largest of the countries. And, you know, just doing business in Korea doesn’t mean a lot, especially as we move towards AI more and more. And because of that, I just want those...⁓ kind of people who are ambitious to really change the world in a significant way, not just build the next chatbot or the next food delivery app, but something that kind of, you know, breaks around and just changes something very significantly. That’s something that I’m really looking for these days.Grace Shao (35:00)That’s really interesting. Do think that has anything to do with your upbringing, just being so internationally exposed?Ethan Cho (35:05)Maybe, actually, yeah. think, this is kind of another personal note. think Asians are really smart in a lot of settings, but we as Eastern Asians, were brought up to be kind of modest and humility was one of our very top priorities as we grew up. And because of that, we tend to be more humble in front of people. And as we know,The Westerners, like this is not a great word maybe, but the Europeans or the Americans are much more aggressive in PR, but we tend to be more careful about that. Back in the days, that was great when we were just living amongst ourselves, but now as we go into the global market, PR is really important and having big ambitions like shoot for the stars, land and the moon is the way to go. But sometimes we just focus on what we have. I think that’s a healthy way of living, but.For entrepreneurship, we really have to dream bigger dreams.Grace Shao (36:02)That’s really interesting. I think it’s some things that I’ve even really noticed within the just generating Chinese founders as well. It’s really different. Like you mentioned coupon. I think the founder was Harvard educated, but he returned to Korea focused only on the cream market, just like the last year. He’s like the JD.com Alibaba’s and the day these are Chinese market businesses. They have global footprint, but there no one’s thinking of them as a international business, right? At the core, they’re Chinese company. But if you look at theEthan Cho (36:18)Yep.Grace Shao (36:28)whether it’s the LLM companies in China right now, or even some of the more consumer facing ones, or even the robotics ones in China, I kind of feel like there’s a shift in generational behavior. Exactly to your point, some of them are less educated than they’re not, but in general, people are not as modest. People are actually more, not in a bad way, but they’re much more open to doing PR for themselves. Well, not just PR, but actually flexing and going more ambitious, going global.like we said, like Kimi and Minimax, whatnot, Jiu-Jitsu, they’re used globally, right? And they’re not shying away from it. I think that’s really interesting. That’s like a phenomenon across East Asia right now. So I think for us to understand, what are some, I guess, misunderstandings or things that foreigners who are trying to invest in Korea often...you know, get wrong or not completely get correctly because, know, obviously there’s so much societal nuances. Well, in South Korea is a country where I find, like you said, it’s not not only not that immigration friendly, but actually in some ways a bit more closed off, Much like East Asia in general, like if you’re not from there, you don’t speak the language, don’t understand formality, especially South Korea has a lot of formalities. It’s really hard to do business, right? So how do what are things that you see that foreigners might be getting wrong that they could do better?Ethan Cho (37:44)Hmm, I think, well, I mean, first of all, think Koreans are just a lot of time. I wouldn’t say everybody, but a lot of Koreans are just shy. They’re, they’re friendly, but they’re shy. That’s kind of our kind of default mode. So, you know, if somebody comes to Korea and nobody wants to talk to you, that’s the norm. But once you try talking to any random Korean person, he or she will definitely help you out. That’s kind of the Korean kind of way. They’re being shy because they want to be polite. That’s kind of an Asian thing, right? So.There’s that. think because Korea has been such a small country, think people, some people think of Korea as just being focused on that very regional kind of market. We’re not, obviously. Like we want to also go global, but we just didn’t have enough chance to actually show off that.I mean, if you look at, for instance, the Koreans working in the States, they can show you what a Korean can do if they’re put in the right setting. So I think if you’re an investor and want to work with a Korean firm, think as long as you put the resources and the human talent in the right settings, they will perform. of course, I can’t guarantee everybody will, but in general, that’s how we’re formulated.I think one interesting factoid that I also always kind of want to emphasize is I think China is also similar to this, but because we have this crazy, crazy education system that’s like overly competitive, although we have been really stressed out throughout our teenage years, that actually made us very, very competitive when we just, you we’re put in the right settings. Like we will strive to become number one in whichever setting that we are put into. So just.like, you know, help us get to the right market and get to the right country or what right settings we will perform. So that’s, think, the expectation that you should kind of have for a lot of Koreans and Asians in general.Grace Shao (39:39)just like whoever can go through the national like university exams, like they have resilience. These buddies don’t like they don’t mess up. So I think on that note, I guess I want to ask what are what should we expect Korea to be exporting that if you’re saying that you want to back companies that are going global, you want to back ambitious internationally minded creates, what should we expect? Because no one expected it. Well, not no one. But a lot of people did not expect China to suddenly be exporting LLMs, right? As like one of their hottest new technology right now. I think a lot of times robotics maybe, EVs maybe were more in the expectation over the last five to 10 years because of the strength and the slow momentum it was gaining, right? But yeah, for Korea, what should we be looking at? Like you said, obviously hardware, chips, there were a lot of synergy there. You’re trying to build on top of that, but beyond that.Ethan Cho (40:29)I think the low hanging fruit or the easy pick is K beauty and K fashion. That’s definitely gonna come in the next ⁓ coming three to five years. I personally think there’s a lot of interesting angle in the Korea defense industry combined with AI because honestly speaking, there’s a lot of, how should I call this? Like confusion around the American diplomacy.policy recently because of all these international tensions. And because Korea has always been at war technically with North Korea, I think there is a lot of advancement in Korea technology wise. We have the best semiconductor in the world. I think we are one of the most flexible countries when it comes to, are you going to use US LLMs versus Chinese LLMs? Like we can do both. think we are.We see the pros and cons there. very flexible there, so we can optimize. I think because of that, the defense industry is not only growing very fast. I think it’s a very good place to kind of experiment the new warfare technology without going into actual war. And because of that, I think the Korean defense industry will benefit a lot from AI evolution.Grace Shao (41:41)I see. It’s an interesting area which I’m not like I’m not familiar with at all. But it’s like kind of like you said, it’s kind of one of those areas where you don’t really hope it being really used, right? ⁓ But it’s definitely a very hot space in terms of VC investment and in the US, especially with Palantir driving over the last couple years. Again, not really to AI, but I kind of want to double click on K beauty and K fashion. Why is it like what what is it that you know, over the last week? SoEthan Cho (41:50)Mm. Yeah.Grace Shao (42:08)My husband and were trying to talk about this very casually that day. We’re like, wow, we live in Hong Kong. In the 80s and 90s, everyone was obsessed with Hong Kong pop stars and Hong Kong movie stars across Asia and then even globally. They had all the kung fu shows and then all the police shows. And then in the early 2000s, we definitely had the Taiwan wave, the Taiwan pop stars coming out of East Asia. And even I was in Canada. I was growing up Canada and people loved Jay Chow, right?Ethan Cho (42:29)Mm.Grace Shao (42:36)Nowadays, obviously, it’s all about Blackpink, right? So how does this move around? Why is it going around? And how does one society kind of, I guess, nurture or incubate a global pop star? Is it tied to geopolitical reasons or economic reasons? Or do you think aesthetics?Ethan Cho (42:58)I don’t know exactly the reason if I knew I wouldn’t be working in V.C. I would be another producer. But I think there’s two reasons that I think is the biggest reasons. One is we have a massive farming system, you know. Koreans train boys and girls in their early teens.Grace Shao (43:04)Yeah.Ethan Cho (43:18)to become the K-pop stars and you have to go through vicious vicious competition to actually get there. So because of that, I think there’s so much talent that’s going through that pipeline, which is a blessing and occurs at the same time to society, obviously, I think so. But there’s that. The second part is I think Korea had a mix of...American culture very early on because of the Korean War and the Korean forces, sorry, American forces staying in Korea. So if you look at Blackpink’s music, because you quoted Blackpink or even BTS, there’s a lot of African-American music, ⁓ like features within embedded in that music line, in the melodies. It’s very some of it is reggae, some of it is very hip hop. And those kind of cultural fragments were embedded very early on because we hadmore exposure to African American or hip hop music versus let’s say China, which didn’t have American troops staying in China. So there was that. Then somebody might ask, what about Japan? They also have a huge American troop there. I think Korea was because, maybe because we were a smaller country, we were more open to actually getting into and using those vibes. And because of that, think.The Korean kind K-pop or K-beauty, K-fashion factory has become a little bit more westernized early on and that kind of made the entrance barrier a bit lower for the American market. That’s kind of my hypothesis.Grace Shao (44:49)Yeah, because if anything, kind of going back to your point on like South Korean and East Asian companies and people don’t really do a lot of marketing and PR, I would say ⁓ Korean cultural export has been extremely successful and has been a really, really strong soft power export. So I want to end the conversation on again, back to AI. What are some things that you think we might have not covered today? You think we’re missing? Like, what are some trends?Or say like if we really spoke again, let’s hope not two years later, but let’s say we spoke two years later, what would be true for you to think of how society has evolved, what Korea has maybe contributed in a global AI supply chain ecosystem, how to understand how you view the future.Ethan Cho (45:32)One thing that we haven’t touched that I’m personally passionate about and interested in is the mental health industry. It’s going to be very different from now versus three to five years down the road. As we know, the fitness industry, physical fitness industry, has become a huge industry ⁓ after the Industrial Revolution because people started to use less and less of their muscles. I think that’s exactly going to happen for our minds and brains.⁓ And because of that, this is not going to be driven by AI, but it’s going to be kind of a side effect or a secondary industry from the AI revolution. To keep everybody healthy, think this is something that we as a society and company as country has to work on. there’s going to be, I don’t think this, I don’t necessarily think of this as a dark scenario. I think as we go to the gym, we can go to this mental gym or something.very on a regular basis to keep ourselves healthy mentally. I think that’s gonna be something very huge. Until now, I think we’ve focused a lot on the hows, like how are we gonna do this? How are we gonna do that? The answer to that has been AI and robotics. There’s gonna be more and more questions about what are we gonna build with this? And after that, there’s definitely gonna be questions about why, why are we doing this? I think that’s not just gonna be philosophical, but it’s gonna be a very practical question.that will lead to a lot of business opportunities. So I think that’s something that we’ll have to question ourselves and answer and discuss on a very regular basis down the road to reach something meaningful either as an entrepreneur or an investor.Grace Shao (47:07)I think that’s really, really meaningful. And I think, you know, we kind of touched on like companion bots and even your you mentioned your son might be even using a companion bot himself. I don’t want to probe on a personal level, but actually on this note, then how do you view that? Like, do you ever fear that he’ll be too dependent on it or, you know, I could be creating a false reality?Ethan Cho (47:27)I think it really depends, right? I know this is not the best answer, but like I’m a big fan of the movie, Her. I think it was a very, very good example of how things can evolve. The ending was kind of sad and happy at the same time, but until then, he was very happy with Samantha. So it seems like there’s definitely a scenario where we can be more happy about the world, be more thankful about the world, thanks to this.Grace Shao (47:33)Mm.Ethan Cho (47:52)maybe emotional buffer that we create with our AI companion. There’s definitely that. But there’s also going to be a downside because the companion will feel real, but it’s not going to be real. So how can we cope with that? It’s going to be something. I still think it’s going to be very similar to the fitness industry just because when we do like, you know, bench presses or, you know, like all these like that pull downs, those are not actual resistances. We’re creating them artificially to strengthen our muscles.So I think our minds should also be strengthened in that way so that we can cope with all these scenarios that we’re not gonna be able to actually experience down the road because we’re gonna live in our own world, which is gonna be safe and creepy at the same time. you know, a lot of factors that will change down the road. So kind of excited and horrified at the same time.Grace Shao (48:42)No, 100%. I think your point on mental health, you use like a general term, but there’s obviously the obvious fear, like what we just talked about, like psychosis and dependency, but there’s also kind of like you mentioned, touched on like, you know, if we don’t really use our brain that way, you don’t really know how to do it anymore. Just kind of like languages, you know, when you move to a country and you don’t use that language for a while, you lose it. Math, I like literally don’t know how to do math anymore. It’s pretty sad. But you know what I mean? Like if these are skills where like you kind of havepush yourself and the gym is something quite, if you think about it, very arbitrarily created for our modern day lifestyle, which obviously didn’t exist even like two generations ago. But yeah, like I think that’s a really interesting take. I don’t know if that’s actually your differentiated view, but you know, I usually always ask one last question to every single guest that comes on the show,what is one differentiated view you hold? So something that might be a bit non-consensus, it could be provocative, it could be not, know, it could be about industry, it could be about life. Honestly, I think what you just said earlier was a bit, it’s quite insightful. It’s something not talked about in the mainstream enough, but if you have another one.Ethan Cho (49:31)differentiate the view. huh. I can make really dangerous comments here. ⁓ but I think,Grace Shao (49:55)No worries.Ethan Cho (49:56)yeah, this is one thing that I always think about. So I think that the creator cannot make something that the creator has not experienced. That is something that I think deeply about, and that is my personal view on the limitations of AI. How we think about AI is to become this everlasting thing that works 24-7, does only good things for humanity. Buthave human beings actually ever experienced that? I don’t think so. And that’s going to be a big question because we’ve never, we don’t know how to work 24 seven. Well, of course we’ve, you know, we’ve done all nighters for sure, but can we actually think of a process that can continuously work 24 seven by thinking, not just operating machinery and also can we think of a kind of standard that is always only helpful to human beings? Like we haven’t really done that.So I mean, I think that’s gonna be a big challenge. Like, however we construct the system or the standards for AI and robotics and all these systems going forward, there’s gonna be a loophole there. And that’s something that we’re gonna have to figure out as a society as a whole. So I think that’s gonna be something that it’s gonna be very interesting down the road.Grace Shao (51:08)Do you kind of, are you kind of alluding to what we’re seeing right now? A lot of people have AI fatigue where they actually make the agents just work 24 seven for them. So essentially the moment they just stops doing a task, they repeat, like they let’s restart it. That’s kind of the work, right?Ethan Cho (51:20)Yeah, I think so.that definitely shows what the problem is. Because we don’t know how to operate these. So I’m facing the same. I don’t use it as much as I used to like a month ago because of that. Because I’m feeling that, this is controlling me, not me controlling that. So there’s this reverse effect. So I think it’s a good thing that a lot of people are already kind of figuring that out. people are kind of.trying to like healthily distance themselves from all these agents. So that’s, I think, a positive sign. But I think as a society as a whole, that there’s going to be more and more things that we’ll have to think about.Grace Shao (51:55)No, I totally agree. And I think there’s certain things. There’s a lot of value in stopping and thinking about the action before the action respoots again. Obviously, there’s certain repetitive work that can be streamlined. so much of accessing knowledge work. mean, this discussion can go another hour, but so much of the whole argument on knowledge work being completely replaced just seems a bit I feel naive for me. Like, I feel like so much of the knowledge work actually requires us.creating things and I don’t know maybe I don’t understand technology well enough so who knows maybe it can create things on its own. Ambanao, I really really want to thank you for yourEthan Cho (52:26)Yeah, that’s another. Yeah,thank you. Thank you, that’s another hour of conversation so we can do it next time.Grace Shao (52:34)Yes, please. Thank you.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Assembled co-founder John Wang on building a AI native support system for enterprises 04.05.2026 43λ

  In this episode, I sit down with John Wang, the co-founder of Assembled, to explore how AI is revolutionizing customer support. Having transitioned from a Stripe engineer to an AI startup founder, John shares his unique insights into the evolution of support tools. We delve into how these tools have shifted from being mere cost centers to becoming strategic assets that enhance customer experiences. John and I discuss the impact of AI on support volumes and staffing, highlighting how integration is reshaping the landscape. He emphasizes the importance of talent density and assembling high-caliber teams to drive success in the tech industry. Through his experiences, John provides practical insights into AI's current capabilities and limitations in support operations.We also explore the strategic considerations for future AI support ecosystems. John shares his thoughts on the role of support in driving revenue and customer satisfaction, and how AI can orchestrate with human support agents to create a seamless experience. His perspective on building high-performing support organizations offers valuable lessons for anyone looking to innovate in this space.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently. Season two will host a series of guests from early-stage investing, as well as builders, founders, and product managers.For more information on the podcast series, see here.To find the previous episodes of Differentiated Understanding, see here.Chapters00:00 The Journey from Stripe to Assembled02:25 Understanding the Importance of Customer Support05:29 Lessons Learned from Stripe10:25 AI in Customer Support: Current State and Future16:04 The Economic Impact of Support Operations18:25 The Role of AI in Transforming Support Jobs24:30 The Future of Support Organizations26:58 Guardrails Against Fraud in AI Support32:42 Navigating the AI Ecosystem38:00 The Value of Long-Term Commitment in CareersAI-generated transcriptGrace Shao (00:00)Hey, John, thank you so much for joining us. I just recorded your bio already. It’s extremely impressive. And you’ve done quite a, you’ve had quite a few different roles now as the co-founder of assembled, right? To start, can you just tell us about your story? Like what inspired you to leave Stripe, you know, go into, you know, right now what you guys are doing, which is a software for people who run customer service support operations. You know, now you guys are pivoting into AI as well, or at least leaning into AI. Tell us about all of this.John Wang (00:28)Yeah, great question. When we, well, when my co-founders and I started, we were all at Stripe. We worked on a bunch of different things at Stripe. And one of the last things that my two other co-founders worked on was a support tool, an internal support tool. And I remember pretty clearly that they were making a bunch of headway. It was really, really cool. And...They had gone to this really, really high up person and product. And this person was basically like, why are you guys wasting your time on this? Like you guys are kind of like, you’ve been at Stripe for so long, you know all these things and you’re doing support. Like I’ve got this really cool Bitcoin project that I would love for you to work on instead. And I remember my co-founder coming to me and being like, hey, like pretty bummed this is what happened.And then I was like, wait, you just saved Stripe, you know, quite a few million dollars, increased customer satisfaction by 40%. And still they don’t understand the value of this. And that’s when we were like, hey, ⁓ there’s something here where there’s a market opportunity. So that’s what got us really, really excited about support. We were doing it at Stripe. We knew it was an undervalued place. We didn’t see any very good tools out there to do support well.And so we decided to go build something really, really great in the support space and just like make transform and elevate support is our mission. Yeah.Grace Shao (01:50)Do you think it was just that stripe was too rich? They were just, and they just didn’t care about saving a couple million dollars? Or do you think it was actually a blind spot for people?John Wang (01:59)I think Stripe was definitely very rich at the time. think it was also a blind, it was a combination, right? Because most people, you think of support, you think of it as just a cost center. And I think recently that started to change in the sense that like, hey, this is actually a really important part of your business. But for a lot of companies, like if you look at FinTech, if you look at like a lot of health tech companies, their entire product is their relationship with their customers.And so support’s actually really, really important for that. And I think a lot of people underappreciated that for quite a while. And now I think people are starting to understand again, hey, if you piss off your customers every time they come and talk to you, that’s not going to be a very good thing. You better be a monopoly. Otherwise, you know, they might not be coming back.Grace Shao (02:46)Yeah, definitely. I think I want to kind of lean into that later in our conversation as well. It’s like people are trying to replace support and customer service AI first. But if anything, it’s not the best experience when you’re frustrated with a product and you keep on getting a robot, right? But I want to kind of talk more about your experience at Stripe. You were there quite early. What do you think it taught you, you know, as a very early employee at such a successful startup now?if even considered still startup and then like what were things that you think you learned there lessons even if soft skills that you kind of took away to to your current role like as a founder.John Wang (03:21)Yeah, it’s a great question. You know, it’s really funny actually. I just met up with someone where, so when I was starting out of college, I had applied for all these jobs. I was able to get a lot of them, except for this one company that I really, really wanted to go to. It was called Meteor Development Group. They built open source software. In college, I had built open source software at Ruby on Rails.I was really big in that community. was like, wow, it’d be awesome to go and make this something I do day to day. And I didn’t get the job. I was really bummed about it. And then I was like, I’ll just fall back on my second here, which is Stripe. And Stripe was the obvious second choice because just the people were really, good. And now like 10 years later, I like think about that and I’m like, the business model is really important.because Meteor was not a good business. Like open source frameworks is not a good business, but Stripe, really boring. Honestly, it’s just like payments. You process payments, you go talk to Visa. You literally have to like, we had a server in the server room that would send like a specific file with specific tabs and spaces in order to get it out to Visa. Really boring. Really, really core infrastructure too.And so like the big overarching thing that I learned was like one, business model is unbelievably important because if you can just make a good product when the kind of like market is there and when there’s a really big need, then this can scale like unbelievably fast. Two was the people. I remember talking actually to a few people, Greg Brockman was maybe the second or third person I talked to.who’s now the co-founder of OpenAI. And I remember just talking to him and being like, wow, this person is so, so smart. This is awesome. And I would talk to kind of like, I would go to the lunchroom and be talking to people at Stripe. that was just, people were talking about all sorts of things. And I think like talent density was a really, really big part of like what made Stripe successful. AndIt wasn’t any one thing over time. was one, Stripe was in a great market. And then two, it iterated really, really fast on a lot of little things over and over and over again. So I thought that was a really good place to learn a lot about like what makes a company great.Grace Shao (05:49)Yeah, I think it’s interesting you’re talking about talent density and a lot of the AI labs I speak to actually also talk about that. But I’m curious, what does it mean when you have really strong talent? Is it like that they are technologically superior, like they can code better? Or does it mean actually that they can think outside the box, they’re more creative, they can pivot faster? Like what does it really mean to have really high caliber talent on your team?John Wang (06:11)I think it depends on what company or like what you’re trying to solve, right? Like talent density for Los Alamos national, like Los Alamos, like building the atomic bomb is like very different than talent density for like Bell Labs, which is very different than talent density at early Stripe, which is very different also than talent density at OpenAI Research. Like I think for Stripe in particular, the type of talent density that was there was really high curiosity.Grace Shao (06:31)Right.John Wang (06:38)really high product thinking, really technical people, and people that could dive deep on certain problems and weren’t afraid to go talk to a bunch of customers. You saw so many conversations about like, how do we make this particular API parameter better for everyone? And like hours and hours and hours of like making sure it was a really, really good product. And people who weren’t afraid to like, you know, take a week of work and just like dump it away because it wasn’t quite there. So it was like,This combination of like, they worked really hard, they’re really smart, and they care a lot about the end result and have a high quality bar. That was Stripe’s version of kind of like talent density. But I think like, you know, if you look at the labs, if you look at different research institutions, maybe it’s just, you know, I don’t know, the raw ability. Yeah. But.Grace Shao (07:26)research capabilities or whatnot, right? No, that makes a lot of sense. Yeah, I wanted to ask you earlier on in our conversation, you said, you know, look, a lot of people overlook support. It’s not that glamorous. People kind of think it’s like a back office thing. But, you know, is that was that your view back then? How does you kind of, I guess, lean into this? And did your perspective or support change over the years? Now you say it’s very important, right? Did you understand the category correctly? Do you think?John Wang (07:52)You know, I think that when we looked at the category, we went at it from like kind of the lens of, Stripe was this company that worked in this unsexy space and did really, really great things. And we thought very similar things about support. It took us a long time to really grock support. And we talked to hundreds and hundreds of different people across different parts of the support stack. AndI think early on, honestly, it was good and bad in certain ways. It was like, we thought we could build a piece of software really quickly that solved everything. Or like, you have that problem, we can build that in two weeks. Not a big deal. And we could solve the specific problems that they had in two weeks. And I remember talking to actually a few people, which was like, the system that you’re trying to do, which is called Workforce Management for Support.that’ll take you seven years. And we’re like, no way. Like we can do this. We can do this so fast. It’s going to be done soon. And now like seven and eight years later, we’re still working on it. We’re still uncovering more and more things. And that was probably the right, you know, that was probably the right call. But also there’s like some importance to naivety, which is like, if we had known that we wouldn’t have started. like we, yeah, like.Grace Shao (09:06)That’s why a lot of people say, yeah, as founders, right?John Wang (09:09)Yeah, so I think it was the right thing to do, which is just like start building stuff.Grace Shao (09:14)It’s amazing. ⁓ Why don’t we pivot into actually understanding your product bit better? So for someone who has never worked in support ops, what is the simplest way to explain to them what Assemble does? Because even between us, we had calls, we had back and forth emails. I was like, John, I don’t understand what you guys do. I’m trying to read through this material. I’ve listened to a few in the interviews. I don’t know what’s happening. Can you just dumb it down for me and explain to me what exactly you guys do?John Wang (09:37)Yeah, for sure. Let’s say you have like 10 people on your support team and you only do email, then you probably just staff them nine to five, right? Like there’s no big deal there. Once you start having a few more people on your support team, let’s say you have hundred people now and you might want to chat to your customers because AI chats, AI chat bots are a really big thing. Then you actually need to start thinking aboutwhen do these chats come in and how many people do I have in order to handle those chats, right? Because like, if you were talking to a chatbot, you’re getting instant responses back and forth, back and forth. And then you’re like, I have to wait 48 hours for the human response after I get handed off. That’s a really bad experience. So the problem is, you you’ve got a bunch of people who are calling in to support, writing in, who are chatting in.and they’re coming in at all different times of the day, they’re calling in for different types of problems, right? You, on the kind of like back end, you have a bunch of people and those people might be able to do different types of things. Like I might be a really good person to handle, you know, where’s my money kind of issues, but I might not be as good at like ⁓ fraud issues, right? Like if you’re having problems with fraud on your account.So there’s a lot of ways in which you can actually put people to the actual incoming tickets. And what our platform does is it tries to match those two things up. if you think about supply and demand, supply is the people that you have and demand is the people, like your customer is asking for questions. And if you don’t match those up well, you’re gonna either...spend way more money than you need to because you’re just going to staff everything way above what you need, or you’re going to have a terrible customer experience because it’s going to take you a really, really long time to get back to people. So it’s really an efficiency play. How do we make it really, really efficient for you to answer questions? In the last few years, we’ve also added AI agents, which is, you know, how do you actually respond instead of just with people, but also with AI togo and answer a chat or answer a phone call directly using AI.Grace Shao (11:50)That’s amazing. I really didn’t know there was so much like science kind of going behind that. I just thought kind of like you’re on a chatbot usually you have to have your frustrating like get me someone, get me someone. I’m one of those people who like no pages pressed zero all the time. I’m like, get me a human. But it makes sense. actually once you can match the talent with like the issue, it can be a lot more efficient in solving the issue and the customer experience will be much better as well. On the AI agent side.What’s the kind of, I guess, consensus right now? Like, are they really actually good at solving issues? Are customers complaining about them? Like, ⁓ how sophisticated are they at this point? He’s like, in my day to day, you know, obviously calling the banks or DHL for pickup or package returns, whatnot. None of those agents are really a pleasant experience, frankly.John Wang (12:35)Yeah, I think this depends pretty drastically on what tools you give these agents access to. I would say that the standard experience right now is fine. It will answer knowledge questions for you. And these can solve anywhere from 30 to 60 % of incoming issues, depending on how many knowledge questions you get.the place where it really is important is when you actually give it access to say your backend database and you can like make a refund or you can look up in order or you can identify why is my what’s going on with this error, right? And that is actually the hard part that prevents most of these banks and airlines and etc agents from being very good is because like thatAccess to data is a thing that they need to actually run and be able to perform actions. And then also the evals for that are really, really hard and not something that you just like launch without really thinking about it. So I’d say it’s in a progressive, like it’s in a progressing state, not at a place where it’s like, this is absolutely solved, but there are also some of our customers who have 90, 95 % of all issues who are able to be completely automated.because they’ve spent the time to give access to all of these systems and spent the time to validate that the agents are performing.Grace Shao (14:00)Very interesting. ⁓ I want to pivot into the to be kind of angle. Who are you guys actually selling to? Like who are the people inside companies that are managing this? Is it the head of support operations? And when they are buying and assessing your product like yours, is it really winning on price? Is it like over, you know, other maybe large softwares? Is it winning on speed and service? Like help us understand essentially how you guys are succeeding winning over customers.John Wang (14:26)Yeah, we generally sell into the head of support. Sometimes that person rolls up into the COO or there’s a head of operations or something like that. But generally there’s some group that is working on unsupport related things and that’s who we sell into usually. I think generally our differentiated, like the way that we actually go and sell this is one, we know all about workforce management, which is like a really, really nitty gritty detail about how youmake your systems really good. And it can save you millions and millions of dollars. Almost actually, and this is one of the things that’s really funny, it’s like using our AI agents versus using our workforce management, we actually see somewhat similar gains across those two. Because to use the AI agents, you’re usually doing it so that you can reduce head count, right?And in order to reduce headcount, you need to know how much can I reduce headcount without hurting my customer experience. And for that, you generally need something like Workforce Management. So what we do is we go in, usually we have Workforce Management helping you understand how your system is set up. And then what the AI agents that we can also bring in is a relatively easy sell becauseour AI agents are really, really connected to kind of like how you staff and when you pull in people. The thing that you were mentioning, which is like, hey, my bank still doesn’t have a very good experience, that’s true of a lot of places. And getting access to information is really hard. So escalating to a human actually happens pretty frequently, sometimes 20, 30, 40 % of the time. So getting to the right human or the rightor figuring out when to escalate to the right human is a really, really important skill to have. If you spend a million dollars a year with me, I should escalate you much more quickly than if you are a free user and you haven’t spent any money for me ever. Similarly, depending on the topic, depending on what kinds of things you’ve already previously talked to me about, I should be able to get to different types of agents and I should be able to have different levels of thresholds.that send me to a human. And I think because we have and handle the workforce management side, our ability to do the handle time and to make sure that you’re getting to the right person is much, better than a lot of our competitors.Grace Shao (16:45)So there’s an unspoken tier system then I guess with customer service as well that we don’t realize. How should we think about the economic importance of support operations? In terms of, we always think of it as like we said, back office support, but how much, do you have any proof that like basically better customer service equals better revenue?John Wang (16:52)There is, and sometimes it’s spoken. But yeah.You know, that’s a good question. should probably have some specific proof here. I guess the best anecdotes I can find are usually the kind of like medium to long-term anecdotes where companies that do not invest in their customer support tend to, you know, regress to the meat, right? Like if you are really trying to bare bones your way through customer support,⁓ Your customers will understand that and it’s not going to affect your revenue right now, but it will likely affect your revenue in 6, 12, 18 months the next time that purchase happens. have seen actually some of our customers, so in our AI agents, we have a configurable setting that’s like, do you want to be containment focused or do you want to be escalation focused? And how good of a customer experience do you want?And we’ve generally seen actually that there’s a strong correlation. Obviously we haven’t run a ⁓ natural experiment or a true A-B test with this because it’s pretty impossible. But you see a general correlation between the customers that spend more money on support, the customers that spend more on trying to have a high quality experience, and the revenue growth of those companies.actually most of the customers that we spend a lot of money on care so much about support that they actually have, you know, executive briefings every week about these, about what’s going on. And they’re the people who have the largest support teams and they’re the people who kind of like make the most, make the most changes with their team. Obviously this is a very biased perspective from our customer, like set of customers, but I think that there’s still something to that where if you spend money and if you want to make ayour support really, really good, that does tend to pay off with customers because they do tend to notice and it makes it easier from a product perspective to paper over all of the things that aren’t so great.Grace Shao (19:06)Yeah, no, totally makes sense. think even as consumers ourselves, we would be likely turned off by certain brands or experiences if the customer service really bad, right? Unless, like you said, they’re a monopoly and there’s nowhere you can go. All right, let’s talk about AI. You kind of touched on that earlier, but the naive view is that AI automate support, you know, a lot, a lot of the conversations right now about, my God, jobs are going be taken, especially the first batch is probably in roles like operationals and customer support roles.Second batch, people are saying are maybe in like more repetitive execution roles like junior consulting roles, a lot of junior training up roles, right? How do you see that? Because at least from where I sit in Hong Kong, a lot of stories are coming out saying markets like India, the Philippines, know, across Southeast Asia where they traditionally served as those telephone call centers or operational centers, they are getting caught. Is that going to be a trend forward?you know, how should we understand this?John Wang (20:02)Yeah. Yeah. I think there’s a few things. There’s like, like with all things, there’s a lot of nuance to this, which is I think your trend on seeing, you know, what we call tier one support, the first line of support who are traditionally humans outsourced. That is a place where we’re seeing a lot of change. And I don’t think that trend is going to slow down. That said, there’s a very interesting other trend thatwe’re seeing, which is that total spend on humans and headcount isn’t necessarily going down by that much. And it’s kind of like Jevin’s paradox where we see a lot of our customers and a lot of customers of other AI users ⁓ who have amazing resolution rates. They’re like answering so many questions, but that’s causing actually, or maybe there’s some correlation here ofthe number of tickets they’re getting and the number of chats they’re getting is like way, way, way higher than before. And I think there’s a few parts to this. One is you see way more ability for your AI agents to answer questions. And so obviously people are going to ask more questions because like, Hey, it used to be really hard for me because I had to literally type out an email to a human, wait a few days and get an answer. And now I can just like get an instant really good answer. Right. So I’m going to try asking more questions.The second thing is, as these companies do better and better, you actually just have this natural induced demand of increasing usage, numbers of people who are asking for support. So the higher amount of support that is automated is also, the general number of how much support is coming in is also very high.And so that actually offsets a very, very large portion of the head count. The head count is changing though. It’s not going to be the typical tier one support where it’s just like, answer an easy question. That is mostly going to go away to AI, think. The types of head count that is coming in are like, know, internal agents, people who are really good, people who can provide white glove support and like...actually go talk to people and provide like a human experience because like our companies still want and really crave giving that experience to people. And that’s just not what the kind of BPO standard really is. So I think it’s changing in the type of what you would see.Grace Shao (22:31)Yeah, I was actually going to ask about that, like, as in when AI agents start resolving more tickets, if we’re just going to see reduction of headcount. And I think you answered him when you once wait, whereas like, yes, in initial stages, but later on, there will be new jobs created, right? Essentially, people who will be managing more critical issues or even managing the agents. I want to understand. So for your company right now, essentially, are you a are you like a middleman between the human agents and the AI agents and becoming the orchestration layer, like you’re providing the service, the training and the orchestration. Like, how do we understand that?John Wang (23:05)Yeah, that’s a great question. So we think of ourselves as how do we get you to the right way to answer your question, right? In our view, there’s kind of three main types of people that can answer a support question. One, it’s AI. Two, it’s a tier one BPO’d outsourced agent. And three, it’s an internal agent who’s like super well-trained and like super, like, you really carrying about, like really trained on customer support. And what we are trying to do is make sure that you get placed at the right area, depending on what kind of issue you have and who you’re talking to and like what is the kind of like a cue that is backing up the set of people who need calls. So for us, what we’re trying to do is really provides you that ability to choose across a bunch of different options. So we don’t actually provide any, like we don’t provide any BPO agents, we don’t provide any internal agents. All we do is provide the software that routes you. And we also provide the software that can do the AI agents, or you can actually plug into a different piece of software if you want to have your own AI agents too.We’re trying to make sure that we are kind of third party and that we are making it really easy for you to optimize your support regardless of what specific providers you use.Grace Shao (24:30)So in your view, what does a well-run support organization look like in, let’s say, three years as AI adoption becomes mainstream or more more mass market?John Wang (24:38)I think you’ll probably want to have all of the different types of support using AI. So voice AI, chat AI, email AI. I think you’ll want to have a lot of nuance between the different types of customers that you have. You can’t generally provide the best level of support for literally everyone. Though this depends on also your customer base, right? Like a consumer customer base versus a super enterprise customer base with 100 very large customers is completely different. But let’s say for a standard company that might have ACVs that are in the, I don’t know, the 100 to couple tens of thousands range, then you’re probably going to have a combination of AI agents and human support. And you might have different tiers of human support, right? Some human support that’s really good at answering support questions and other tiers of human support, which is like, you’re just managing the agent. I think the other thing that’ll happen a lot is you’re gonna start to see more like, supporting agents acting in a simulation where right now, like the kind of typical flow is like a supporting agent gets a ticket and they answer it and it goes back. I think as the agents get more like, get more and more training data, get access to more information, really they’re only gonna come to humans for escalations. And similar to how Waymo works, if you’ve ever taken a Waymo, it’s a great experience, you’re like driving, driving, driving, and sometimes you get kicked out and a human operator in the Philippines is like, hey, I need to move you around this truck, right? And similar to support, That’s probably what’s going to happen. A human operator is going to come in and be like, hey, I can give you a refund right here. And then what’s going to happen is the AI agents are going to train on that, right? They’re going to like learn and get better. And you’re going to be able to use that whenever you have an interruption to understand like, why did I have this interruption? How do I make my model better for the future? And then you’ve got your closed loop. So I think in the future, you’re going to see much more of that happening than people who are just like, coming in and their job is to solve as many tickets as possible. I think the change is gonna be like, okay, people are gonna start to need to provide the best possible response in that particular instance so that the models can train on that and be as good as you are.Grace Shao (26:58)actually just on that, do you think then we’ll see more and more fraudulent activity or people trying to exploit that? like if say you know the models trained on, I say this one buzzword or one keyword and it triggers like refund. What if I just go on the call like on the phone all the time, just to be like keyword, keyword, you know, and then like how do we prevent something like that? Or do you guys kind of get involved in that building this guardrails as well?John Wang (27:21)Yeah, no, that is a age old question. think like, wouldn’t say there is going to be necessarily more or less of that, but I think like, it’s kind of like the cat and mouse game of like, everyone has always been doing that. And so like, and the methods always change every, every few months. I think the methods will change every few months here too. Our AI agents have a lot of guardrails put in place to automatically detect that. And we also have kind of like post-hoc guardrails which are like scanning through our logs and trying to identify situations where that might have happened. And we’re also training on those examples, right? So I think, yes, people will definitely start to exploit this and be like, hey, how do I get a refund faster? But there’s a ton of guardrails that you can put in place. For example, each account, you can have one or two, have like refunds without looking until that actually gets flagged and it needs to go to a human or.You can set good policies, for example, like, you know, if it is within policy of 30 to 40 days after purchase, like automatic refund, otherwise, you know, flag it and do something with it. So there’s a ton of stuff that you can do to actually like reduce the possibility of that. And I do think that it will end up being cat and mouse game like over and over again, as people get more sophisticated.Grace Shao (28:39)Right, right. And they’ll start using AI to trick AI. That’s what’s scary, right? So as we talked about, different gender standing the podcast does not have to interview anyone related to China or Asia, but we do have kind of an Asia angle to a lot of how we view the world. So my question for you really is because you’re like out in San Fran and like your company actually has no sales in China or anything. But I actually had a curious question. How does SFJohn Wang (28:43)Totally, yes.Grace Shao (29:05)as a whole, the startup ecosystem kind of view the current rise of a lot of Chinese AI. And have you guys yourself or your peers, you know, tried to use Chinese open source models over the years? Is there any view on the open source models given that, you know, you previously said you were very involved in open source and I think it’s part of your philosophical belief as well, right? So just kind of like the high level vibes.John Wang (29:28)Yeah, our vibes might be different than at the model, like the Frontier Labs, honestly. Our vibes, we love the Chinese open source models because it adds more competition. And I think the open source models are actually very, very good. I think from my friends at OpenAI Anthropic, they don’t like it quite as much because it’s competition. But for us, we have no allegiance really to any of the Frontier Labs.or any of the models that are out there, we want to provide the best possible experience to our customers at the best possible price. And that has meant, you know, over the years, like making changes in our models, making updates and to figure out what is that frontier of cost or performance. The Chinese models tend to perform really, really well on that, especiallyGrace Shao (30:12)Mm-hmm.John Wang (30:19)kind of like the latest series of models, we’ve actually spent a lot of time in the last six months kind of like pulling out a lot of our tokens. We have tens of billions of tokens per day. And a lot of it now goes to models like Quen or Kimi. And like that has actually started to really, really increase over time, mostly because you can find to them, you can do RL on them, you can...have better latency on them, you can run them on your own hardware. There’s just like so much more stuff that you can do with it. And also, you know, the cost performance latency trade off is really, really good. Now, most of our most of the like the strategy we take is actually one where we try to understand the use case and the problem and what type of model is necessary for that. So for kind of like the main model that’s actually answering questions. We’re actually usually using a frontier model for that. But actually the majority of our tokens come from out of secondary processes, processes like detecting if I need to escalate, detecting if there’s a fraud here, detecting if there’s an adversarial intent, making updates to large swathes of data in batch, like all this other stuff where you really don’t need frontier level intelligence and where if you have a a well-tuned prompt and an open source model or an open source model plus a fine-tuned model, you can get at or better in terms of frontier performance. We’ve really seen that and we’ve actually been able to save millions on our token costs in just the last two or three months by being very smart about how we use our models. And we’ve also seen a 15 to 20 % increase in quality.⁓ Just because like when you go and you have evals, you can make things much, much better more quickly with these open source models.Grace Shao (32:14)Yeah, I think that’s like the general sense I kind of get from a lot of startups, right? In a known day, it’s like, you guys are obviously more cost conscious. What is the best price to get to what you need? And there’s like a tier system where how you use the models, you might not use the most frontier models for everything. I think that makes a lot of sense, business sense, especially. Is there anything you would like to share with us that we haven’t touched on in terms of, just the overall AI ecosystem, any thoughts on, you know, where we’re going with this AI agentic push right now?⁓ you know, are we really going to see that, you a giant moment, like just kind of some high level thoughts.John Wang (32:49)Yeah, that’s a good question. Recently, I’ve been thinking a lot about Opus 4.7, which got launched a few days ago. And it’s actually kind of similar to what we were just talking about in terms of this price for performance ratio. And it seems like, based on my usage, based on our evals, based on other people’s usage on the coding side, that it’s a better model, but it is also more expensive.than before. like, you’re really it’s literally like a trade off in terms of dollars and intelligence. And it’s really interesting because, you know, a year ago, every model would just be like, this is strictly better, and it’s probably cheaper, and you’re to get more context and like, everything’s better. And you could basically just bet that you’re just going to like get better models across the board. And now actually, you’re just like kind of moving from this part of the like the frontier curve to the other part of the frontier curve without actually shifting the entire curve. And that’s happening with a few more model releases. You still see general increases in the frontier, but it’s less stark every single model release that you see that. And so I think it’s just an interesting area to look at because when you get into that world.Gross margins has become really important. Gross margins for ourselves as a startup, but also gross margins for Anthropic and OpenAI. One of the funny things that I’ve seen, just talking to people who are working at Anthropic and OpenAI, and also people who are trying to invest in those companies, gross margins are actually incredibly important. One of the OpenAI right now is becoming a much more...hand investment than before. And like, it used to be like six months ago, it’s like, you have it, you have shares of OpenAI, like, how do I get in? Now it’s completely different with, you have shares of Anthropic, how do I get in? And I think part of that’s because like, OpenAI wants to spend $100 billion on infrastructure. And Anthropic is a lot more measured in the way that they’re spending money. And I think gross margins actually do matter a lot right now. And that’s where I think actuallyChinese open source models are making a big difference because just at the end of the day, you still have to make money. And if you’re losing money on a per token basis, that’s really bad because if you go to infinity, you lose infinity money. And if you make money per token, great. Ramp usage up as high as you can.Grace Shao (35:03)Yeah. It’s just so crazy how the sentiment shifts like so every three months I feel like and then to your point like whenever I speak to investors like oh my god I got my hands on some anthropic shares and last three months earlier. Oh my god I got my hands on opening I like it’s just like and like oh no one would invest in opening right now like I don’t want to do that like people just completely go like black and white on these things it’s pretty crazy how the pendulum swings I do have a question actually on the infrastructure side doesn’t it actually make sense for open AI to eventually own their infrastructure because otherwise they have to becomecontinuously constantly pay the hyperscalers for all the infrastructure like so in the grand like scheme wouldn’t it make sense? I mean although obviously how much you’re spending is like absolutely crazy.John Wang (35:55)I think it actually does. And I think that’s like part of the problem, which is like, you know, if you think about what their compute costs are, I think actually doing all of these things that they were doing makes perfect sense. And it makes especially perfect sense if you have investors who are willing to bankroll this. But it’s almost like the, ⁓ what’s that paradox? It’s like the St. Petersburg paradox, something like that, where it’s like, you keep doing,your expected value is infinity and you keep doubling your money basically, but at some point you need to not double your money because you don’t have enough money.Grace Shao (36:32)That’s such a mo- I’m like, I’m still confused when you’re saying, go back. You keep on doubling your money.John Wang (36:36)Sorry, So I think the I think it’s like Let me let me look this up st. Petersburg paradox is Okay, it’s a coin flipping game and You start at two dollars and with every tails you double the pot and you can basically decide to like take your money at any time, right? and so you you’re doubling exponentially as you go up andIf you compute the expected value, you should basically just like, keep going forever because your expected value is like infinite, right? Like because the doubling of the pot is better than kind of like what your losses are. You just got to, you got to run. And I think OpenAI is in this St. Petersburg paradox where it’s like, well, in theory, double everything, keep going. But in practice, you don’t have enough money.Grace Shao (37:17)Yeah, I see what you mean.John Wang (37:25)and resources to be able to do that. I think that’s actually what’s happening is like, there’s not enough money in the world, not enough investors with liquid cash who are willing to invest in a business as big as OpenAI while the gains and the returns are still, yeah, having improvements. So I think it’s both rational, but also, you know, actually practically very hard to make what they’re doing.Grace Shao (37:41)haven’t been proven. Yeah. totally. Okay, I want to ask you one last question, which I ask every single guest. What is one differentiated view you have? It could be on your own sector, industry, life.John Wang (38:00)man, have a really like, I have one that like is very controversial. I don’t know if I should talk through that one. ⁓Grace Shao (38:07)You’re get doxxed and to hate it after this.Okay, tell me that one after, I wanna hear it.John Wang (38:17)Yeah, yeah, Let’s see. Like... I would say, I don’t know if this is differentiated now in the market or not, but the thing that I’ve been thinking about recently is that you should stay somewhere long enough where you see your mistakes through. And I think it’s like slightly differentiated right now, because like you’ve got in Silicon Valley, at least you’ve got people who are jumping between big labs, who are jumping between different startups where it’s like, Hey, I can make the next, you know, $5 million.by going to this next thing. And there’s just a whole huge amount of opportunity and there’s like a ton of opportunity costs to staying somewhere for a long time. And at the same time, think like long-term staying somewhere for a long time is actually one of the best things that you can do for your own learning. And it gives you that a better shot to make like the long-term massive gains that you could have like $5 million.is amazing for someone. But if you want to build your own startup, if you really want to like change everything, if you jump around between companies every year or two, like you’re probably not actually going to learn a lot. And you’re probably not in the position to make really hard decisions and then have to see those hard decisions through and then, you know, be able to learn and see that feedback from those hard decisions. Especially if you’re jumped likeEspecially if you’re like at OpenAI and you’re like, no, investors don’t want this anymore. You jump ship to Anthropic. That’s like, you know, I don’t think you’re going to get that, the learning that you really need to get.Grace Shao (39:46)Yeah, yeah. actually agree with that. think I also took some time and like experience to realize that because when we’re all young, like you’re really excited, right? It’s like, this looks cool. That looks cool. this person hates me. hate that person. Like you take everything very personally and then, you know, we’ve all heard these stories from peers, even ourselves. But what is the threshold though? Because then the other side of the argument is that like you see people who’ve been in a job for like a decade and clearly they’re frankly not.moving up in a very corporate structure way or even intellectually growing or even, you know, excited about their job anymore. You know, the joke is like you get the like, okay, this sounds on PC, but you know, like the 45 year old VP that’s been a VP for the last 15 years at banks, we have a lot of these. So what happens? Like when is it best for them to actually maybe jump or some say, that was like a lifestyle decision where they want to take it easy because they have some more time for kids. Fine.But taking that kind of considerate way, wouldn’t it sometimes be better that you jump to try something new to take risks?John Wang (40:50)I think if you are in a place where you’re unhappy with, so I will caveat this with, have to, you should only stay if you’re excited about what you’re doing and you’re learning continuously and you’re surrounded by great people. If those three things aren’t true, yeah, it’s really hard to fly.Grace Shao (41:05)Which is so hard to find. you were very lucky at Stripe, right? Like you said, you were just surrounded by very high caliber, high agency people, but not everyone can get all those things at the same time. ⁓ But no, that’s great. Thank you so much, Sean. ⁓ I had a lovely time chatting with you. I still wanna follow up on what was the unspoken differentiative you later. All right, thank you.John Wang (41:17)Yeah. ⁓ Let’s do it. Let’s do it. Thanks, guys. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Matt Sheehan on China’s AI Policies: Employment, Anxiety, Safety, and State Priorities 27.04.2026 1ώ 1λ

  Today, I’m joined by Matt Sheehan who writes this insightful newsletter. Matt is a senior fellow in the Asia Program at the Carnegie Endowment for International Peace. He researches China’s AI ecosystem, Chinese tech policy, and how technology shapes the country’s political economy.Matt lived and worked in China from 2010 to 2016 and later led China tech research at the Paulson Institute’s MacroPolo. He’s the author of The Transpacific Experiment. He speaks Mandarin, and he turns complex policy into plain English.In this episode, he helps us understand China’s AI governance, about how Beijing is thinking through the social and political consequences of rapid AI adoption. We focus especially on a shift that became more visible in early 2025: rising concern inside China’s policy community about AI’s impact on jobs, worker anxiety, and social stability.Matt explains why China’s AI labor question is different from the Western debate. We also discuss how the Chinese government is trying to balance support for technological progress with the need to manage public anxiety, clarify labor rules, and avoid social instability as AI becomes more deeply embedded in the economy.He broke down the myths, explained the jargon, and the regulatory bodies in China. Our conversation started slow, but it became very, very heavy, what they call 干货满满 substance heavy. Also, a shoutout to Nathan Lambert’s work in helping us better understand the open-source ecosystem and Rui Ma’s for helping us understand investing in China AI!Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently. Season two will host a series of guests from early-stage investing, as well as builders, founders, and product managers.For more information on the podcast series, see here.To find the previous episodes of Differentiated Understanding, see here.Chapters00:00 Introduction to AI Policy in China03:10 Matt Sheehan’s Journey into Chinese Tech Policy05:55 Shifting Perspectives on AI and Labor09:02 Public Concerns Over Job Security and Government Responses15:09 Education and AI: Preparing for the Future17:50 Regulatory Landscape of AI in China34:00 Navigating China’s AI Regulatory Landscape40:58 Misconceptions About Chinese AI and Government Funding43:57 Understanding AI Safety and Security in China52:03 Global AI Governance: Cooperation or Parallel Paths?AI-generated transcript Grace Shao (00:01)Hi Matt, thank you so much for joining us today. I’m so, so happy to finally have you on the pod for people who are listening. We’ve been trying to make this happen for like six months, but between us, there are like three little children running around with a bunch of viruses and have just not been able to make this happen. I’m really excited. ⁓ A few months ago, what really caught my attention about your work again is that you shared something on WeChat saying you were dissecting the new Chinese AI safety paper, like the big national one. ⁓like verbatim in Chinese. And I was like, wow, this is extremely impressive. It’s not an easy task. I commend you for doing that. So I really wanted you to help us understand the nuances of the AI policy world, especially how people are perceiving AI in China. I think there’s more more interest in how China’s governing AI ⁓ while we were hearing the backdrop of how the Chinese government is trying to push on AI diffusion, right? And then on top of all of this, like where areas where China’s AI governance seem to be leading, because in many ways it seems likeChina’s AI regulators are much faster to respond to how fast technology is evolving. But to start, we would love to hear about your personal story. Tell us about how you ended up studying China, studying Chinese tech policy. We met in Beijing years ago, maybe a decade ago. ⁓ Yeah, so tell us about that.Matt Sheehan (01:11)Sure.Yeah.Yeah, sure. Sort of stumbled into China stuff. I hadn’t taken Chinese or really knew anything about China until about halfway through college when I ended up getting a summer job in Beijing. I was just kind of like instantly fascinated and knew I wanted to move back there after I graduated. So took a little bit of Chinese my senior year, moved to Xi’an, taught English, kind of followed what at the time was a very like typicalknow, trajectory of like, go there, teach English and then go study Chinese at university and then get a job and get a slightly better job. And eventually I was able to kind of wiggle my way into journalism. And so I was a China correspondent for a publication called The World Post at the time. And that took me up. was there from 2010 to 2016. So kind of like the hinge period before and after she came to power. Pretty interesting thing to see. AndWhen I moved back to California in 2016, I started working on a book about China-California ties. I’m from California and this was like the period of kind of explosion in cross-border investment and Chinese students come into California in the Silicon Valley-China relationship getting even more like twisted and complicated. China-Hollywood. So I wrote a book about that and as I was doing it, the kind of the tech section, the China-Silicon Valley, China-U.S. tech connections kept growing bigger and bigger and I ended upworking a little bit with Kai-Fu Lee on his book, AI Superpowers, which was kind of my turn from like, it was like all things China, China, California, China, Silicon Valley, China AI. And since 2017, I’ve been working almost exclusively on AI issues in China. Maybe the first three years of that, like 2017 to 2020, was very focused on comparative capabilities. This was kind of a period right after the National AI Plan in China when there’s a big explosion in activity. And I think...This is kind of was like the first time America kind of got freaked out about Chinese AI capabilities. And so I spent a few years being like, okay, let’s try to like ground these assessments in some data. Let’s get like an actual grounded sense of where the countries are with each other. ⁓ And then starting in 2021, I sort of turned into focusing on Chinese AI governance, Chinese AI regulations. That’s when they first started rolling out their regulations or recommendation algorithms and sort of deep fakes. And I was kind of making a bet that I thinkIf China continues to be at or near the frontier of AI, then how they choose to regulate it domestically is going to have huge implications for China’s own ecosystem. And then it’s going to really ripple out internationally on safety, security, growth, all this stuff. So I spent the last, now it’s like five years, ⁓ just deep in the weeds of Chinese AI policy and regulation.Grace Shao (04:04)Oh, great. I think I definitely want to double click on all the algorithm security and the kind of, you coined the answer versus what’s called security and versus what’s the other Chinese word? Yes, yes.Matt Sheehan (04:17)Anshun safety security. ⁓Jeff Ding was talking about this very early on, but yeah, it’s a, it’s a constant thing that we have to negotiate for people who don’t know it’s the word Chinese, the Chinese word Anshun ⁓ means both safety and security. whenever you’re kind of translating documents on this front, you have to know, you talking about AI safety, which is kind of a different thing versus AI security, right. I think both you and Jeff definitely are some of the more nuanced scholars I follow. And I do want to kind of double click on that later on. But to start, I think I want to talk about something that’s top of mind for a lot of people. You just wrote a piece that you said was not super serious. It was just your scattered thinking put together on subset. I thought it was very well written about the growing anxiety around potential job losses. ⁓your perspective is that you know there are more and more people voicing this kind of concern and I wanted to hear a perspective on that and I kind of wanted to share a bit of my different share my different perspective on this and what I’m hearing on the ground and kind of have a conversation around that as well. Yeah why don’t you start with sharing like what you found yeahMatt Sheehan (05:22)Yeah, sounds great. Yeah. So, I’m not an AI and labor person. That’s not been my focus for a long time, but I’ve been monitoring it for a long time and just lightly. And starting around, I guess it was early 2025, I just started to hear a lot more out of the Chinese policy community about worries about AI’s impact on labor and jobs. And this was kind of a surprise to me because ⁓ just a little bit prior to this, say early 2024,I had, I sometimes ⁓ in my job, I run these kind of like informal surveys or almost like a, what do call it, focus group of American and Chinese AI policy people and asking them like, you how would you rank these different risks? How concerned are you about ⁓ job risks versus privacy versus military AI? And we have both sides that like rank the risks and then talk about the results. And when I ran one of these in early 2024, it was very striking that the Chinese sideI think it was at the time we had seven different risks and the Chinese side ranked labor impacts a second to last as six out of seven. And so my sort of baseline was like, OK, for a variety of reasons, this isn’t really too on the radar of China’s ⁓ policy community or wider policy community. And then starting around early 2025, some of those same people who I had been talking to about this before had really changed their thinking. They were saying that there was a big change in thinking withinChina, maybe especially within China’s of policy and government circles, but then I think also a little bit wider. And so that sort of sparked my curiosity. And for the past, now it’s like over a year, I’ve been just sort of tracking when does this AI and labor question show up in state media? When does it show up in kind of the online discourse? When does it show up in policy documents? And sort of the TLDR is like, I think this is really, really ramped up a lot. Over the past 18 months. It’s been maybe the single biggest change in how China perceives different sort of risks as it relates to AI. And I think I’m looking forward to sort of discussing how maybe like the policy world or the government’s perception of this differs from ordinary people or certain categories of people. ⁓ But I think it, from my perspective, it’s sort of it’s infused into both. think there’s been a fair amount of public concern.the policy community picks up on that and they want to both respond to like the actual problem, know, actual job losses, but they also really want to respond to people worrying about job losses. That’s kind of maybe the thing that actually made me write this piece now was discovering an interesting piece in state media. I think it was in science and technology daily. That was the headline was like ⁓ AI must be controllable, but people’s ⁓ anxiety about AI must also be controlled.And it was all about how sort of OpenClaw has triggered a lot of anxiety and a lot of people about, are they going to get replaced? You need to be building your own AI agent in order to not be left behind. And they’re sort of trying to tamp down those concerns in a few ways. So that’s what sparked the piece itself.Grace Shao (08:33)Yeah, I think definitely what you saw and you wrote about is like definitely kind of playing out in the China AI policy ecosystem that I see as well. And I think for sure, the open-claw frenzy have kind of opened the eyes to lot of the even average people what AI could potentially do. However, I guess my argument, not against it, but it’s just like, you know, we kind of cite each other’s work on subset. But my point was kind of saying, you know, this is a reflection of a relatively elite group of people end of the day, because the knowledge work economy in China end of the day is only like only 30 % of the workforce actually are the knowledge working economy. And end of the day, even though it’s 30%, because China has such a huge population, the mass, the sheer scale, it feels really large. However, I want to bring it back to the idea that like anyone who’s lived in China understands that the government’s like top top priority really is about social stability which leads to what they call social harmony, right? And I just think that, you know, the rising anxiety of job control a lot of times maybe is because there’s a fear of if there’s a lot of disruption to jobs then people will lead to social unrest which obviously gets a bit more sensitive but you know, a lot of what they do comes from that I guess thinking so I agree with you top down definitely have to understand what’s happening with technology and how advanced AI has become in 18 months.⁓ have given them kind of, I guess even fear mongered a little bit internally, right? ⁓ But the nuance here is that I think ⁓ the rest of the 70 % of the Chinese workforce actually don’t work in anything structured that we know. I think even probably even 80%, you know, people in China, they most of them are actually like, you know, service providers ⁓ from, you know, rural areas and urban areas. A lot of people work in factories, even the entrepreneurs, right? They run like say hospitals, clinics, factories, bottle cleaning, like factories, whatever, right? ⁓ Car logistic rental businesses, these people aren’t actually ⁓ trained in the way that maybe the West by default think they are. They actually just run it from a grassroots way and they don’t have very, very streamlined processes. They don’t have documentation. They don’t run like what we think a corporate has run. So in that sense, I think it’s very hard for AI to replace any of their workflow.because it’s actually not a, we can’t really provide context and a lot of things, business is done is through one C, is through a wink, through a look, through a gesture, through, you know. So a lot of that, I think, in fact, will be harder to replace than even maybe some of the more mature businesses in the West where there are structured processes and everything. So that’s kind of my, I guess, a more nuance, I think, push on that. Yeah, wonder what you think of it.Matt Sheehan (11:20)Yeah. Lots of thoughts. And I think the sort of the fundamental distinction that you’re pointing at is very valid in that like, you know, companies in the West, in the United States, they’ve been like big companies have been running sort of digital databases for decades. They have like decades worth of data. They have pretty advanced like enterprise software. It’s just a much more I want to say something like a bias, but it’s a little more like put together sort of official structured⁓ technological backend and not just technological, but like a process backend. Whereas in China, it’s just things have developed really quickly. A lot of it’s on the fly. lot of it is, know, enterprise software is just not, there’s not really a market for that in China in the same way. It’s a lot of stuff is pirated or they’re just not digitized in the same way. And it’s actually very interesting. This is in kind of the early days of the like China, US who’s ahead. ⁓ know, a lot of the debate focused on data.Matt Sheehan (12:18)And there was this idea of China has a billion people, so it must be this huge advantage in data. But my pushback on that was always kind of what you’re arguing, is like the US actually has very structured data, and it’s owned by corporations, it’s deployed by them, they’re already doing type of sort of lower end market intelligence type stuff. ⁓ So I think that that sort of backdrop is very real. think ⁓ maybe from there I’d like differentiate out to potential risks or debates. And it’s kind of what you were pointing out as well. There’s like the actual question of how many jobs are going to be impacted. How many people are, what is it going to do to people’s wages? What’s it going to do to aggregate employment? And then there’s the question of like, ⁓ how do people think about that? What are the fears due to sort of the Chinese social stability, even if the things haven’t manifested. So I think separating those out, definitely the government is... ⁓their sort of initial response is related to public worry about this. So in the piece, I detailed the way that ⁓ sort of a robo taxi incident in Wuhan was in many ways the spark that really like ramped up the government thinking on this. And this is something that I heard from a couple of different Chinese policy people who both pointed to this incident, which I had totally missed at the time and wasn’t like major international news, but that had a big impact. And basically what it was is thatBaidu was rolling out its sort of fully autonomous robot taxis throughout the city of Wuhan. There was this kind of like public letter, ⁓ open letter released by a taxi company that was kind of railing against, you know, both ride hailing platforms and autonomous vehicles as, you know, stealing the iron rice bowl or just smashing the rice bowl of taxi drivers and of companies. And even though it was a kind of like a small thing, a couple of days later, a Baidu taxi actually hit pedestrian, I don’t think they were seriously injured, but it kind of fed into this overall, a big kind of online reaction and discussion about like, what’s going on with AI? Is it going to take people’s jobs? And it, it’s one of those things, it’s funny to explain to people because it sounds like nothing, but it did lead to a pretty significant ⁓ imprint on the way that the Chinese government is thinking about it. So it was coming from, in many ways, public discussion of it. Like the discussion was happening online. This discussion might be happening among, you know, elites.chronically online people. ⁓ But it’s something that the government definitely picks up on. So I that’s one element. They’re worried about the worries and they want to, like among their sort of policy reactions in a ways, thus recently has been sort of directing platforms to say like, you kind of need to tamp down these articles or these viral videos that are telling everybody, like if you don’t adopt open claw, you’re going to be left behind. There’s been this kind of rash, both in China and here in the US of like,you know, kind of like fear mongering people into clicking and taking your course on building agents or just subscribing, whatever. And one thing the government is doing is telling you like, chill on that. Like, don’t be putting that narrative out there. ⁓ And so that’s part of this kind of like public opinion management thing. In terms of the actual impact on jobs and who will it hit?It’s a huge open question that I personally have gone back and forth on for years. I first kind of did a deep dive on this way back in 2017 when everything was still so speculative. At the time, was pretty not worried in part because of the reason you described it. I’m like, there’s just a lot of friction. There’s just so much friction in this economy. And just because an AI system can theoretically execute a test doesn’t mean it’s taking a person’s job. And for me personally, my thinking on this has changed a lot in the lastyear to 18 months, mostly because of how capable agents have proven to be. I expected agents to essentially be hitting a lot more roadblocks while they’re being deployed online. They haven’t been. They’ve been operating much smoother, or just they’re more relentless, and they can break through these bottlenecks. ⁓ It’s interesting that in China, the inciting incident was not about white-collar workers. It was about taxi drivers. ⁓I don’t know this as a fact, but if I had to guess, I would guess that a much larger portion of the Chinese population’s job is driving a car or driving a scooter or something like that. And that maybe that’s a vulnerability that they might face depending on how self-driving vehicles roll out or delivery robots and stuff like that. The knowledge workers, yeah, it’s such a messy and unclear thing, but I think the government has at least started to take it seriously because it’s not, their policy responses are not just this like,public opinion management stuff. They’re also talking about, ⁓ like one of the more interesting pieces that I highlighted is, and maybe the most concrete thing they’ve done so far is, ⁓ according to Chinese labor law, there’s sort of reasons why you can and cannot fire a person. There’s like legitimate and illegal reasons to fire someone. And when someone is fired and they object, this gets taken to like a labor law mediation ⁓ body that’s under the Ministry of Human Resources and...forget what the second part social ⁓ security. ⁓ Yeah, Ren Li, Ziyuan, Shouhui Bao Zhang. Yeah, that’s what it is. ⁓ And in the last year, one of the things that really made a lot of made kind of a big splash is that those mediation bodies declared and it was echoed in like the biggest state media that saying that you replace someone with AI that AI can now do this person’s job is not a legitimate reason to fire someone and those people have to be reinstituted into their jobs.That’s a concrete policy thing that’s actually directed quite clearly at like actual impacts. ⁓ Is it going to work? I really don’t know. It might just be a little bit of friction and, you know, maybe China’s kind of doing what it always does, which is like, we’ll figure this out. We’ll kind of muddle through this. We’ll put some friction here. We’ll grow a little more here. But I think the concerns are real, whether they bear out, ⁓ whether they hit faster in the United States or China, which country is better positioned to sort of roll out a more redistributive welfare system. think these are all open questions, but I think the concerns at least are real.Grace Shao (18:24)Yeah, and I think you hit something that I feel like it’s being kind of missed in the headlines, which is the government actually cares more about the general mass, which are the people who are driving the scooters and the like the DDS cars more than the knowledge workers, which is kind of different from the Western kind of conversation right now, where a lot of the whether it’s fact, frankly, the power that can lobby and the power that the voice that have the voice are all really concentrated in.the white collar elite jobs that are very much concentrated in Silicon Valley and whatnot, right? And I think it really reminds me of the time when, you know, during the Hulianmang Shidae, like the internet era, you know, like the big tech only really got clamped down when the average consumers felt like they were really being pushed to R-Shrine Egypt 2-1. So that’s when they had to add the monopoly to probes. And then soon after, only maybe two years after the probes happened, there was a common prosperity rollout, which basically all the big tech in some capacity had to like showcase that they had a CSR aspect to them. I think this is something that we don’t really see in the US as much with all the big techs, because it’s kind of like they’re doing what they need to do. They have their profit driven interests. And then of course, everyone has a CSR, but it’s not really allegiance to the government CSR mission. It’s more like, we believe in ESG. We believe in climate. Amazon is going to have some like carbon footprint reduction plan, right? Whereas like the common prosperity thing rolled out. ⁓you know, it kind of died on its own, like no one really talks about it anymore. However, during that phase, when it did get rolled out, it was like an understanding where, OK, if the government wants the, frankly, the poor or the middle lower class to feel protected, then you as a very large ⁓ moneymaker in the economy need to showcase that you are somehow ⁓ part of this kind of support. So I wonder how this will play out for the big tech in China when, the job protection policies really get rolled out in practice, like what you’ve mentioned. And obviously they can’t really say, you’re being replaced by AI. At least there’s that superficial guardrail there, I think.Matt Sheehan (20:24)Yeah, I think the sort of the political economy of these questions is going to be super interesting in both countries. know, essentially like business in many ways, it inverts the sort of technological impacts on employment that have been around for so long. Normally, like greater technologies integrated into the workforce, it hits sort of people working maybe low end manufacturing jobs or jobs that would be considered sort of repetitive and, you know, quote unquote, low skilled jobs, even if they’re not. ⁓And, you know, in the United States, we’ve seen like three, four decades of this. And the people who are concerned about that basically didn’t get hurt because they are not, like you say, part of these influential classes. AI is going to, yeah, in the United States is going to be very different. This is going to be the first time that you have the way that my sort of mental model for it is like, if you’re a senator and you have kids or nieces and nephews or your friends, kids like what, what are their problems?And like how close do those feel to you? And you know, if you’re a 60 year old senator and you’ve got like a 23 year old niece and she just graduated from college and she got a degree in, you know, something that’s like is legitimately employable normally like in marketing or something like that, and those jobs just aren’t there, that’s just going to feel very close to home for people in power in the U.S. in the ways that it hasn’t felt in past waves of technology impacts. In China, I...partially agree with what you’re saying, but I think there is also going to be a significant element of this sort of the same dynamic as the United States. mean, yes, common, you know, Xi, common prosperity. He’s focused a lot on sort of eliminating extreme poverty and, you know, the CCP, it’s in its bones that like the rural, the working class are in many ways kind of the long term support base of them. But I mean, also if you look back at Chinese history, like a lot of the biggest and for the government most dangerous protest movements came out of elite schools, came out of students at elite schools who ⁓ either couldn’t find jobs or were facing inflation issues or had, for ⁓ ideological reasons. I think that stuff does hit close to home. think some of those same dynamics, if you’re a deputy director at the National Development and Reform Commission.your family, the people that are close to you, are going to be the type of knowledge workers that are going to be impacted by this. And I think that just can’t help but kind of like compress in on the thinking on this. ⁓ You know, how AIs can... Yeah, yeah, and like how...Grace Shao (22:58)Yeah.everything becomes personal in the end. Like in the end, it’s likepolitics is still personal. Yeah. Sorry.Matt Sheehan (23:08)Politicsis personal and yeah, mean, like cities are where social instability is the most dangerous. Like cities are where people gather and you can have potentially dangerous incidents. These are the people who are very online and are sort of sparking or leading the conversation as much as that can be controlled and manipulated via censorship regimes or public opinion guidance. Like these people are gonna be vocal. yeah, I think if I was at the CSPI, I’d be concerned aboutGrace Shao (23:39)I want to kind of go into on education. Like you kind of touched on it, right? Like, you know, there’s been draft rules about children’s interaction with AI in China as well. There seems to be more guidance and obviously concerns around that ⁓ from at least from the top down ⁓ about their mental state, their dependency, or even what constitutes as an AI companion, how we should draw the line on that. We know likefamously a couple years ago, China installed this rule where like, you know, kids under 16 cannot actually play online games on their own without the parents consent. However, that you know, there’s obviously loopholes in practice. But again, it goes back to there are, you know, rules and laws in place to try to protect minors. How do you view all of this? ⁓ Because in the with the backdrop of China trying really hard to diffuse AI into the real economy. And then there is this pushback like you just mentioned onconcerns about AI taking jobs. I feel like there’s also almost like a ironic kind of contradiction happening where, you know, Tiger moms are like, okay, now we don’t need to learn math, we know how to learn AI. And Tiger moms are like saying, how do we optimize getting into, I don’t know, Harvard with AI’s help? And how do we get AI into the education, education, ASAP? I mean, honestly, we don’t even know what the education system might look like in like two decades.from our kids, but at this point, seems like there is like embrace. I don’t know. How do feel about that?Matt Sheehan (25:13)Yeah, a couple strands there. One just on the sort of regulatory side, like this is a long term strand in Chinese tech policy and tech regulation. They always put a pretty heavy emphasis on like how are kids using technology. They have, ⁓ they sort of mandated having like a minors mode on various ⁓ apps. ⁓ This is the regulation I think that you’re referring to as the newly passed. ⁓I translate as anthropomorphic AI. That’s the word that’s the official translation. ⁓ So it basically means AI that, you know, behaves like a human. This could include AI companions that are, you know, literally like a character pretending to be your friend. They could also include, you know, the way that people interact with chat, GBT or Kimmy or whatever, you know, the phenomenon of having AI boyfriends and girlfriends and all this stuff. So there was a new regulation on this that was just finalized, I think, last week andIt has some protections for everybody, specifically around ⁓ self-harm, addiction, and stuff like that. But it has really ramped up protections for minors and for elderly users. So there’s all these kind of specific add-on requirements. For ⁓ minors, it involves permission from parents. Parents can review at least some. They can set limitations on how the child uses the system. They can review.conversation that might have got toned down a little bit in the new version. ⁓ But I think, this is many ways it’s the same concern that surfaces in the US and elsewhere, like California just passed a just passed last year, passed a similar regulation on AI chat bots that I think also had believe it had extra protections in there for kids ⁓ on the education side of things. I guess there’s a couple of things. One, there’s just like the yeah, you say the tiger mom’s like this is ⁓It’s a booming industry of like, I’m going to teach your, you know, four year old AI so that they can use it because this is going to be how they get a job and how they get into school and how they get a job. ⁓ you know, a lot of it is bogus. Maybe most of it is bogus, but it’s very attractive to parents who have grown up in a really, really cutthroat competitive education system where you’re looking for every single edge that you can find.And so that’s, that’s a piece of it on the, from the policy side, they have both sort of AI and education policies, AI plus education policies that they’re pushing in a bunch of ways. I have some friends who are working with teachers over there who are described to me pretty like sophisticated and interesting ways. The teacher that are using AI to lesson plan, to create like really interesting games that keep the kids engaged and learning stuff like that. So you have those, and then on the labor.the labor side of things, they’re also viewing AI, they’re also viewing education as something of an antidote ⁓ to AI fuel job disruption. This is in the, I in the five year version, it is in the five year plan, it’s in the AI plus plan, it’s in a few other places where they say, we’re really gonna prioritize lifelong education. So maybe you used to be an accountant, you lost that job and ⁓ you’re gonna retrain as something else. ⁓which I think is a good, you know, it’s a good attitude to have. If you’re a person, you should always, I’m always trying to learn, you know, books. Um, think they’re great, but I don’t know if at a totally like a, you know, macro population, I know if you’re going to get 500 million people to be constantly staying one step ahead of AI in terms of what jobs it can do now. I mean, a lot of the things that we would have told you go back like two, three years and say, what jobs are going to be disrupted by AI? A lot of the recommendations would have been totally backwards.People would have thought that, ⁓ coding jobs are great. jobs involving creativity, ⁓ illustration, ⁓ stuff like that. AI can’t do those things. It can’t be creative in that way. And it’s like, that’s actually kind of what it’s best at now in some ways. I mean, you can argue about the level of creativity, but like generation of content, generation of images, videos, language. So I’d say it’s a piece of the Chinese sort of response on labor concerns.a fad, but maybe like a ⁓ useful fad within like the sort of education industry. But I’m a little bit skeptical of this as like ⁓ an actual antidote to the disruption that I at least imagine is coming.Grace Shao (29:49)Yeah, it’ll be really hard to be like upskilling, like re-skilling like hundreds of millions of people. Like, it’s just, you don’t even have the capacity to do so if there’s actually mass disruption. alright. ⁓Matt Sheehan (30:00)mean, this was always the response, in the United States on coal miners. We’re going to teach them to code. everybody, all these manufacturing workers, we’ll offer like a job retraining program. I’m like, ⁓ maybe, yeah.Grace Shao (30:05)Yeah.I’ll take generations. I’ll take generations for things to shift, know, resources to shift, people’s mentality shift, you know, for a while, like, when many, when remember the first wave of like, a basic rural kids no longer wanted to work in factories and wanted to go to urban cities, there was a surplus essentially service providers and then like everyone eventually became a DD driver or a food delivery man and thenNow we’re seeing a reshuffle in that population again, where people want to move back to their rural, cities. so I think based on how society is evolving, opportunities will arise without even us realizing anything, hopefully in the best case scenario, where people will find opportunities to reskill. ⁓ But I want to talk about something that’s a bitGrace Shao (30:57)I guess not heavy, but actually not many people understand, even including myself. So you really look at the government and the policy structure of ⁓ China’s regulators in the cybersecurity space and whatnot. ⁓ There’s so many players. There’s the CAC, then there’s the NDRC, there’s the MMIT. I can keep on naming acronyms, but can you give us a really, really quick high level understanding of who’s regulating whom? ⁓how do they actually work with each other and are their KPIs aligned before we get into more about how you know, how China’s policy is shaping the technology and AI ecosystem.Matt Sheehan (31:36)Yeah, maybe I’ll do it. ⁓ I’ll introduce a couple of the players and I’ll do it somewhat chronologically in terms of like when have they become important or rise and fall in importance. So AI policy, like the first really big policy document was the 2017 National AI Plan. It was released by the State Council, effectively sort of China’s cabinet, sort of the highest level of government. But ⁓ people who are in the know say that that was largely sort of drafted and pushed by the Ministry of Science and Technology. So this isreally like the policy wave of like 2017 to 2020 more or less. And it’s the Ministry of Science and Technology and it’s also the Ministry of Industry and Information Technology, MIIT. So these are really the organizations whose job it is to promote science, promote innovation, and MIIT is more of like the industrial applications of the technology. So they were kind of in the driver’s seat in that period of time. They were the most relevant actors. They were the ones who were driving real activity.starting in 2020, 2021, you had the CAC, the cyberspace administration of China, really like come to the center and become the most important actor in AI policy. The CAC, it’s basically the internet regulator. It was created in 2014. It was largely created to kind of like get the Chinese internet under control from a sort of political content ideology perspective. They’re connected to the Ministry of Prop, or the propaganda department.publicity, as they say now. ⁓ So from 2021 through 2023, the CAC was the one rolling out these binding regulations on recommendation algorithms, on deepfakes, on generative AI. And these are the regulations that actually force companies to do things. They actually force companies to register their models, to do pre-deployment testing, at this point to label AI-generated images in different contexts.There’s for that 2017 to 2020. It’s kind like the go-go period. Let’s just like push this industry forward. You have the Ministry of Science Technology, MIIT. And then from 2021 to 2023, it’s really the CAC. This corresponds roughly with the tech crackdown of 2020 through the end of 2022. That was a period when the CAC, the CAC is kind of at least historically, it’s kind like the bad cop of tech policy. They’re the ones who are like telling companies like come in and drink tea and we’ll tell you what you’re doing wrong or, finding companies in different ways. Cyberspace Administration of China, yeah, CAC. ⁓ Some people call it CAC. ⁓ And then sort of one of the more significant changes from 2023 to now is the rise of the NDRC, the National Development and Reform Commission, Chinese Fagawei. ⁓ And they are a macroeconomic regulator. They are like what grew out of the sort of state planning apparatus. And they’reGrace Shao (34:01)This is a cyberspace administration of China, right?Matt Sheehan (34:30)really powerful, they’re kind of a super, super ministry within the bureaucracy, but they’re not sort of directly, there aren’t that many direct connections to AI. They deal with, they deal a lot with money. They have money to give out for projects that funnels into compute projects and stuff like that. ⁓ But they wouldn’t be like who you would think of as the go-to AI regulator. What I was told and what I feel pretty confident ⁓ is what happened is that in some time in, I think, 2023, maybe mid to late 2023 and then into 2024, the top leadership in China essentially said, hey, we need a little more balance in our AI policy. The last three years it’s been led by the CAC. They’re kind of a bad cop. They’re really focused on controlling the technology, controlling the sort of output, the content, the ideology from it. And that is important. That’s kind of their first priority. But we need to rebalance this a little bit. We need to move out of our total tech crackdown era. And now we realize like our economy isn’t doing great.We realized we’re behind the US after CHAT GPT came out, and we need to balance this out. And so they empowered the NDRC to be a of a coordinator across AI policy, someone who is intended to take the input from the various ministries, from Ministry of Science and Technology, MIIT, CAC, and to try to make it little more coherent and balanced. And so that’s kind of the role that they have played for the past few years. The details of how that works out areshrouded in secrecy, you know, you hear little tidbits here and there. But there have been like visible manifestations of it. They had not, when they released these regulations, usually there’s a sort of a lead regulator on it or a lead policy document person on it. And then various other ministries, they co-sign it and they’re like listed below. NDRC hadn’t been on any regulations prior to 2023. And then starting in 2023, they were listed second as like the second sort of most important organ.policy body on these things. essentially we have this kind of like 2017 to 2020 is this like go-go period. Let’s diffuse. Let’s push the technology. Let’s push innovation. 2020 to 2023 is this more constrictive crackdown. Let’s build the regulatory infrastructure for things. And then 2023 to today is just like, let’s balance this out. Let’s not be purely focused on the content and ideology concerns. Let’s also be thinking about development. Let’s be thinking about employment. The NDRC is actually allegedly one of the groups that is very concerned about the employment impacts. you know, tons more details that I will love to go in on, but maybe that’s a starting point.Grace Shao (37:06)No, I think it’s super, super helpful. I just understand the nuances of like what their actual KPIs even are and like, you know, who does what, how they work together. I think that’s really helpful for lot of listeners and even investors who are trying to follow the space and just confused by acronyms. But help me understand now, like you say that 2023 to now essentially is in the same kind of era. However, I feel like at least from the capital market perspective, you know, the last year might have seen a bit of ashift again, you know, it was a bit of a let’s go AI, big tech AI, all the labs, let’s go, let’s go IPO. Then obviously the deals, some of the deals didn’t come through, some of the IPOs didn’t come through. ⁓ There seems like you even said people are being told to tamper down their excitement a little bit. Is that aligned with what’s happening with the policy side of things? Or is that actually more a reflection of just, frankly, you know, the AI space not being that exciting right now, you know, since the Gentic ⁓ kind of breakthrough. We’ve not seen more consumer and breakthrough. Also, there’s a lot of talk about, you know, there’s no obvious proof ROI on all the spending from all the big tech right now. Help me understand all that, I guess.Matt Sheehan (38:15)Sure. Yeah. When I was breaking down those errors, is largely its policy, but it’s already kind of like government attitude towards it. It’s like which, you know, they’re always in some ways swinging back and forth, going back and forth on the seesaw between, you know, control development, control development. And that 2023 to now being one era is sort of in that sense. It’s the period of rebalancing more towards development. There’s tons of sort of wiggles in that process andthings they’re pushing more and retreating on. But from a positive perspective, that’s the overlay. ⁓ In terms of like the capital markets, investments, I mean, I think this is kind of at least for people in the United States, it’s kind of like the one of the most misunderstood things about the Chinese AI ecosystem is that it is really like cash constrained, that it is not like the United States where, you know, open AI is just like sucking in.the tens of billions of dollars from a huge variety of investors are just spending huge capital outlays, which people talk about, is it a bubble? Is this going to come back to bite them? That’s an open question. But in China, you don’t have the concern about that bubble because there just is not the same level of infusion of cash. when a couple of the companies did IPO recently, Z.ai, formerly Jerpool and Minimax IPO in Hong Kong, and I think I’m notGrace Shao (39:28)100%.Matt Sheehan (39:39)really an IPO guy. think the IPOs were like modestly successful, but the valuations are just, yeah, the valuations are, yeah, not even close. And ⁓ it reflects a lot of things, but it largely reflects like a funding environment, a business environment, a macro economic environment, and the general sort of attitude towards risk investment. think I was just reading something that ⁓ Ray Ma from ⁓ TechBuzz.Grace Shao (39:42)So that’s six to eight billion dollars. The valuation is tiny compared to American peers.Matt Sheehan (40:06)China was writing on this. She’s always very good on these topics. yeah, it’s just people kind of assume that there’s like infinite money in China. They’re like, yeah, the government, whenever they want to, they just like turn on the taps and then, you know, it’s like, no, that’s not how it works. And like the VC ecosystem is much smaller, much more new and immature. And so it’s a different story.Grace Shao (40:28)on that note you know I was just in SF like last month and I met with quite a lot of investors and people’s kind of I guess misunderstanding was often twofold. One is exactly your point, people are just like oh China’s so rich the government just gives money all these AI companies are backed by the Chinese government I was like 100 % no first of all like there’s some other issues happening in the background but like the government doesn’t even are you know it’s kind of cash constraint and not even that much right now second all these companies are definitely not being backed by the government in any sense in factMost of don’t want to take municipal governments or provincial government money because you get kind of tied into, you know, what we’re seeing is, you you get forced into working with the government and it constrains your profitability and commercial goals. On the other hand, another really big misconception was, I thought quite funny was that people often ask, was open-claw frenzy because the Chinese, average Chinese consumer or user were really, really concerned about privacy issues. So they wanted everything on edge.I was like, hmm, like again, it’s kind of like just not, a major conversation people have. Like I think I hate to generalize, but I think because of how the internet ecosystem is in China, people by default have kind of ceded to not thinking too much about privacy or personal data issues as much. So that definitely isn’t. So I kind of want to bring the conversation that this, you know, likeWhat are some biggest misconceptions you think people have and how do we help them understand and bridge that gap a little bit better?Matt Sheehan (41:59)Yeah, I think yeah some of the stuff that you point out is correct like If you’re if you’re a if you’re a startup if you’re like a small medium company You I’ve talked to these people they’re like actually like we do not want to take government money if we can avoid it not just because we get kind of in mesh but like Entrepreneurs are legitimately afraid that if they take government money and then their company doesn’t work out and they lose the government’s money like they could end up like on the hook like in jail thislegitimate fear that it was stated to me by someone. like, you know, is that happening to entrepreneurs everywhere? No, but it’s like you don’t. ⁓ The government. It does a certain amount of sort of VC-esque investing, but there’s not really that VC mentality of like high risk, high reward. Like we know that most of this is going to go under. It’s kind of local governments at least have been trained on like real estate investment, which is like 10 percent, 10 percent, 10 percent every year.And this idea that most of these companies that you invest in are going to fail is not really ⁓ deeply embedded there. I do think some of the companies do rely on government funding in different ways. ⁓ mean, Z.ai, Drupal, one of their biggest, maybe their biggest single revenue stream is from ⁓ building custom models, but custom applications for ⁓ state-owned enterprises, local government, stuff like that.Matt Sheehan (43:28)When you listen to them in interviews, they’re like, it’s not that big. Maybe it’s 40 % or something like that. But it’s a significant revenue. It’s part of their business model. So there’s that type of a connection to government. With DeepSeek, that’s a company that’s kind of quite mysterious. And we don’t know exactly where all their money comes from. Is it all earned? I think the government got more hands on with them in the sort of aftermath of the DeepSeek moment. You had reports about the government taking passports away frompeople who worked there to make sure like you guys stay local ⁓ or the government was like vetting investors was another this is reporting the information. ⁓ But the idea that like these companies are just they just have the kind of like the hose of government money just flowing in at all times and therefore they don’t have to think about anything else is just not it’s just not real. ⁓ They’re they’re much more constrained cash constrained. ⁓terms of like trying to misconception on Chinese AI regulation, AI policy, this is like my, you know, much of my job is like first getting across like, the trend does actually like seriously regulate the technology. And then, you know, the next layer being like, it’s not all people think, you know, it’s an authoritarian system. She didn’t think he must just kind of like sit down and just like write the regulation. So like nothing matters except what he thinks. And we don’t know what he thinks. It’s like, no, like he doesn’t. There’s this actually very complicated and sophisticated policy ecosystem of,legal scholars and the companies are doing their lobbying and their thought leadership and, you know, they’re responding to public outcry over things. And I think that’s a, know, you can get this across to people, but it’s certainly not the people’s default mental model of how China works on policy is ⁓ just does not reflect the kind of sophistication in this zone. And it’s somewhat understandable. Like there are policy areas where Xi Jinping just like makes a decision and that’s.that’s where things are going. Like I think we saw a lot of this in the kind of 2020 to 2022 era. But as an AI policy, COVID, AI policy, it’s not that way. certainly things are not, people are not gonna like, you know, actively push things that are totally against the will of the top leaders, but they are within the constraints of like,Matt Sheehan (45:52)the direction of travel, what the CCP is good with, what she wants to do within that kind of very wide lens. It’s really individual people, scholars, bureaucrats, companies that are filling in all the details on this. And it’s a very sophisticated system because they’ve just had a lot of ⁓ had a lot of bites at the apple. They have like passed, I don’t know, eight, nine different A.I. already.The regulators at the CAC have been getting documentation from AI companies for three, four years. They’ve been building evaluations. been like, and they kind of, got their reps in with AI policy and it leads to a more sophisticated ecosystem.Grace Shao (46:33)⁓ yeah, so the last kind of section I want to focus on is just getting into the nitty gritty about, you know, the policy and the security and safety kind of side of things we touched on in the beginning of our conversation. you are one of the few in the West, I think, and talk about the nuance of the word, which you just explained, it’s security, but also safety. ⁓ help us understand.how to interpret that when we read about that. It actually even helps us understand a little bit of what’s happening in the West. Like, I feel like there’s the governance people, the security people, the safety people, but from someone who might not be in that ecosystem, people are conflating it a little bit. And I just want to understand, you know, how do we understand each of their objectives, again, KPIs, or even their goals?Matt Sheehan (47:20)Yeah, yeah, basically, it’s really complicated. It’s very context dependent and it’s always changing. ⁓ I think maybe the first key thing to understand here is like the very particular meaning of AI safety in the West like that. The West AI safety ⁓ largely refers to kind of a specific camp ⁓ of AI development and policy people that are, you know, believe that AI is going to achieve human and superhuman capabilities.And this could pose like serious, maybe catastrophic risks to people. like, that’s a somewhat coherent community in the United States that has a certain amount of power. Their power kind of ebbs and flows depending on things. But like when you say AI safety in Washington, D.C., it means one quite specific thing. ⁓ In China, that community, it has started to emerge, but it’s much newer. It’s much more recent. It doesn’t have the deep roots that it has in the West.And ⁓ the way that the word is used in policy documents is both confusing and has changed over time. So a lot of times when ⁓ just to kind of put a little color on the terminology, Anquan, when ⁓ when you’re talking about cybersecurity in Chinese, you say Wang Luo Anquan. So it’s like network Anquan, network security and cybersecurity means something very different fromAI safety from super powerful AI systems posing risks. And so there’s one sort of category of mistakes, which is to be very naive and to read all the Chinese policy documents. And every time they say a word that’s translated as safety to believe, wow, they’re talking about AI safety, they really, really care about this. That is a very naive and incorrect reading of things. ⁓ But in the past, I would say, 18 months, two years,you have seen a pretty significant uptick in the way that people sort of in and around the system and to a certain extent in and around the companies, their level of attention to what we would call in the West, AI safety to these more kind of large scale, potentially catastrophic risks from powerful AI. I’d say this is, there’s like sort of levels and degrees of this. There’s people talking about this. There’s it showing up in government documents in one way or another.And then there’s actually implementing this either through like binding regulations or through sort of companies doing their own testing and evaluation to try to their own sort of safety research and their own safety testing. I’d say what we’ve seen so far is a large increase in rhetoric, a large increase in sort of awareness within the policy community about these safety issues. We’ve seen it to start to show up in more significant documents. I think the one you’re referring to early on that I was working working on analyzing is calledThey call in Chinese the AI safety and governance framework 2.0, which is in many ways put out by some organizations underneath the CAC, the internet regulator. And it’s kind of ⁓ their attempt to diagram and do an initial discussion of how they see different risks from AI ⁓ and how are they going to mitigate these risks. Oftentimes they’re focused on technical standards as a mitigation. And there was a AI safety governance framework 1.0 in 2010.fall of 2024 and there was a 2.0 in fall of 2025. And just between those two documents, you can see real increase in the frequency and to a certain extent, the sophistication of the discussion around these risks in China. I’d say it’s pretty significantly below the sort of the AI safety discussion in the United States, but it’s on the radar. will counter, they’re like, okay, that’s great that they’re talking about it, but are what, youAre they just trying to trick us? Are they trying to make us believe they believe in safety? Are they saying it but not doing it? And ⁓ they’re like, we have not seen much in the way of like, we certainly have not seen like concrete binding regulations that sort of implement safeguards on this front. And in terms of what the companies are doing, it’s quite opaque, but ⁓ we don’t think that they’re doing the, I’d say.pretty confident they’re not doing nearly the level of sophistication or intensity of safety testing as you see at places like OpenAI and Anthropic. To me, this seems somewhat normal. This is kind of a process. Chinese companies have been behind. The government has perceived itself as being behind. When you’re behind the frontier, you’re not as worried about frontier risks as you’re like other people are going to get to those first and we need to catch up. ⁓ So I see this as kind of like a long-term process. And I think that the sort of increase in discussion about this isyou encouraging if you’re concerned about these issues. But it’s a really don’t want to be just kind of reading the documents and say every time we see Anquan being like, wow, China cares about AI safety. Look at all this stuff. It’s much more ⁓ nuanced and evolving, ⁓ evolving quickly, I would say.Grace Shao (52:21)I know like when we spoke a couple months ago, just catching up, you were saying a big part of your job is also trying to help, you know, bring the two sides together. Obviously, you know, ⁓ it’s been challenging given the geopolitical backdrop, but how do you think the global AI governance space can work together? ⁓ Are we going to see, you know, kind of the two world superpowers and two super AI powers, ⁓ you know, guide?in different directions or do you think there are certain issues where they need to come together and they will come together and are coming together? ⁓ For example, to your point on safety issues around protecting humanity, protecting children, are these things that you are seeing collaboration?Matt Sheehan (53:08)Yeah, I’m kind of ⁓ both an optimist and a pessimist on this front in that, like I said, I have very, very low expectations for the United States and China to work together on anything. I have very, very low expectations of any type of a binding agreement or some type of detente where we both shake hands and kumbaya and we’re both going to be very safe with AI and we agree and it’s great. I just don’t expect that. ⁓So in that way, I’m pessimistic. I think attempts to try to sort of preemptively create these global governance structures that are going to bind both of the countries in advance so we never reach these dangerous thresholds. ⁓ That’s just not where I’m putting my bets. I think it’s good. We need to make all kinds of bets on this front, and it’s good that people are working on this, but that’s not where I’m putting my bets. Where I’m putting my bets is on a much morelimited kind of narrow bore, but I think potentially highly effective form of ⁓ engagement. wouldn’t even say cooperation. I wouldn’t even necessarily say coordination. ⁓ my sort of the term, my mental model for it is something I call AI safety in parallel, which is that like the two ecosystems are going to be moving somewhat in parallel. They’re both going to be pushing the technology forward. They’re both going to be working through safety issues from a technical perspective, from a policy perspective.And as we kind of move forward in parallel, we’re not going to be telling each other what to do. And we’re not going to be like, okay, I’ll do the safety thing because you are. You told me you’re going to do it, so I’m going to do it. We’re not like sort of moving in lockstep on this, moving in parallel. And we need to have these touch points. We need to have touch points where the two sides develop some form of mutual understanding of what the other side is doing. They understand how other side is thinking about the issues. They understand how they’re perceiving these risks. That’s one of the reasons I do the risk ranking.stuff and in some cases trying to share best practices, explain kind of explain what we’re doing and why we’re doing it and have the Chinese side explain what they’re doing and why they’re doing it and then where possible share good ideas that we think are sort of uniformly good in the U.S. and China. If we think that we have a policy intervention maybe it’s around ⁓ certain types of pre-deployment testing. ⁓ It’s good to communicate that.to the Chinese side. And it’s good to have the Chinese side communicate some of ⁓ the reasons and the sort of the specifics of what they’re doing on these fronts. We’re not here to just like trust each other. I think a lot of people are very worried that the Chinese side is going to, is they’re going to trick us. They’re going to say they’re doing it and they’re not, which is like legitimate concern. know, that’s, this is high stakes like geopolitics and powerful technology. So you don’t take anybody’s word for it. But when you have these conversations in talking with people, you can get a,a sense of their level of sophistication when talking about the issue. If someone is talking about AI safety and they’re like, yes, humanity first, protect the humans, control the machines, that’s our policy. And it’s like, okay, is there anything more to that? They don’t have more than you kind of know that they’re actually not really thinking about it. But if you can get into a more ⁓ deeply engaged discussion, you can see like, actually, yeah, they’re working through these problems themselves. You can see it in the way that they’re.discussing it. You can see it when they talk about their specific regulatory mechanisms. You can see sort of the connection between sort of action and outcome or thinking and action. And so my model for this is like, we’re not going to agree on things. We’re not going to sort of trust each other. But there are ways that we can both be moving forward at the same time and comparing notes, checking in, getting a sense of what the other side is thinking and doing that I think could contribute to safety in a meaningful way.Grace Shao (57:02)I think that’s fair and I think the word you kept on using trust is quite interesting because I feel like whenever I speak to people in the industry, ⁓ there’s just such a lack of trust even within whether you want to say countries or communities or beliefs and value systems and a very, very optimistic, naive way, I really hope that there can be a bit more consensus on certain things like that need to be protected in practice, like such as children, right? And how we go ahead with that. ⁓ But okay, I don’t want to end on a super somber note or anything, but... ⁓ The takeaway is trust nobody. That was...Matt Sheehan (57:34)It’s optimistic in a way. think this can’t... When you do see... Well, trust nobody, buttalk and see if you can share some good ideas along the way. think there is real... ⁓ I’ve seen some real sort of traction from these type of things and I think it’s limited. We’re not going to get some kind of hard guarantee that China is going to be perfectly safe or we’re both going to...Matt Sheehan (58:04)do the right thing. But within with those low expectations, with those kind of pessimistic expectations, there are there’s progress that can be made.Grace Shao (58:13)⁓ I do want to ask one question that’s kind of been happening around right now. That’s been kind of happening like the whole idea of Chinese open models seem to take a little bit of a sidetrack and starting to kind of only release their most frontier related models and close weights. ⁓ Obviously from a very like, you know, capital perspective, where I study it is I feel like it’s a lot of it is because they need to see our eye. They cannot keep doing this because they’re not making money. API sales not enough to, you know, sustain the kind oflong-term ⁓ business as well as research costs. Are you seeing anything from the policy side? Like do you think there’s been a policy shift? That’s also kind of why I asked earlier if this year somehow, know, last year there was a public embrace by the government saying we should open source our technology. Has there been a shift?Matt Sheehan (59:04)So ⁓ I’ve seen sort of little tidbits around ⁓ sort of public policy concern about open weight models, but not enough that I would call it a shift. ⁓ In that document, the AI safety governance framework 2.0, it was interesting because it was the first time that there was a fair amount of ink spent on potential risks from open source models. The risks they were primarily talking about wereessentially if there are vulnerabilities in these models in some way, either maliciously inserted or just a vulnerability mistake, those could proliferate throughout the ecosystem because you have all these downstream models and that could lead to impacts. There’s a little bit of a mention of like, maybe open models will be used by criminals and stuff like that. I certainly don’t see this shift in specifically Alibaba strategy asMatt Sheehan (1:00:04)in reaction to a significant policy shift. mean, it kind of makes, yeah, corporate decision. It makes if you’re going to be spending tens of billions of dollars building models or at least hundreds of millions, billions, billions of models, giving it away for free is a. That’s a choice. I think there’s reasons why it’s advantageous for China to do that, or at least it was for a stage like it was going to be pretty hard.Grace Shao (1:00:09)corporate decision then.Matt Sheehan (1:00:34)to get people around the world to kind of believe in ⁓ Chinese models if they were only going to be able to access them through API. ⁓ You know, a lot of American companies that are, you know, deploying ⁓ Quan, Alibaba’s model, I don’t think they’d be doing that if they, I don’t think they would have at least made that leap initially if they had to sign a contract with Alibaba and they believe that maybe their data was going back to China or, you know, the model was more of a black box relative to them. So maybe the open model wave was a very good period of publicity. ⁓that might pass with time, but I don’t know. I think it’s being more nuanced. I’m not the expert on this. Nathan Lambert, runs the interconnects sub stack, and then Kevin Shue, who runs the interconnected sub stack, are much, much better and more sophisticated on this, and they have good writing on the ecosystem. But from a policy perspective, I haven’t seen a shift. I think it just kind of makes sense.Grace Shao (1:01:35)Yeah. Nathan, I actually spoke about this ⁓ a week ago, and I think both of us kind of feel like it’s really more of a capital and business constraint that’s really driving this. ⁓ But I just wanted to hear from you if maybe there was some kind of a top down initiative as to but it doesn’t seem like it. Right. I have one last question for you, which is a question I ask every single guest that comes on. ⁓ What is one differentiated view you hold? And actually, I try not to limit the question to just about AI. However, most people do want to talk about that.Matt Sheehan (1:02:02)Mm.⁓ Differentiated view. ⁓I mean, is... I don’t know where most people fall on this, but I think one thing I’ve been thinking about lately is just language learning and what’s going to happen with language learning in the era of simultaneous really good translation either through hardware, devices, or software. And I’ve been reflecting, why have I been spending 16 years or more learning Chinese and just in the...Matt Sheehan (1:02:40)mud of trying to learn and remember this language. And I guess my maybe differentiated view on this is I think it still is really important. And I, I waver, you know, I don’t want to just be like, ⁓ you know, justifying to myself why I’ve spent all this time. And like, I, don’t know that I would tell like a 16 year old, like I want you to invest, you know, 10,000 hours on a language when, when you can have it all translated for you, but they’rethere just is something quite meaningfully different about reading these policy documents, about listening to people and hearing the original language and just knowing how that language is used in all these different contexts that gets flattened by translation. I use machine translation all the time. I throw documents into it to get either a first draft or to get something I can share with people. But I guess my view is that I really thinkWe need people to keep like actually learning Americans to actually keep learning Chinese. And I also just think it’s so much fun. just, was earlier today, I was thinking it was like, why is this, it’s been such a, like a joy in many ways, extremely painful, but kind of a joy to like really struggle with a language over time. And so that’s my take.Grace Shao (1:03:59)No, I actually agree with you and I think if anything, I’ve thought about this a lot. So I’m raising a trilingual child by nature because we speak English at home, our parents speak Mandarin, the environment she lives in, they live in speaks Cantonese, right? And I think to your point, in many ways I’m like, wow, it’s actually, it would be so easy for them to travel the world and communicate with people. But the reason why I pushed them to learn the language is really to communicate, to understand a culture and the people and more.nuanced way, even for myself, like my parents pushed me to learn Mandarin. It like to your point, it’s so painful. But the ability to speak to my mother in her native tongue and understand her is so much more complex and you appreciate much more when you’re older, even though my parents speak English, obviously, but when we were young, we would speak English to him. As I got older, I actually enjoyed speaking to them in Chinese much more because you hear aboutMatt Sheehan (1:04:55)Hmm.Grace Shao (1:04:57)It’s also your personality changes, right? Like you kind of get to the core of who they are in their native language and their native way of expressions. So I think for sure, I agree for certain languages, there’s still such value, if anything, even more so to understand a human connection, human connectivity. then for pragmatic reasons, like I took two years of German, I remember nothing. I probably wouldn’t do that again to myself, especially in my class. a bunch of third gen German kids where they spoke the language at home but they can get away with saying they were doing beginner’s German, you know? But yeah, so I appreciate that. Thank you so much Matt, thank you for your time, I really appreciate it, we finally got together to do this episode.Matt Sheehan (1:05:34)Yeah, thanks for having me. That was fun. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Tencent's QClaw goes global, aims to serve the average consumer user, with PM Shuyu Zhang 21.04.2026 45λ

  Amid Anthropic’s success with coding products, many AI labs and companies have also tried to lean into that vertical. OpenAI has stepped back from courting consumers and shut down its video model division, Sora. Alibaba, meanwhile, has more recently begun releasing closed-weight proprietary models and is reportedly pushing the Qwen team to find clearer paths to monetization. The Chinese tech giant has also launched Qoder, a Cursor-like product under the Alibaba umbrella, which we interviewed last year.But despite all this, Tencent remains notably committed to the mass consumer market. The OpenClaw frenzy has already led to five different Clawbot-style products emerging across its ecosystem. Joining me today is Shuyu Zhang, Senior Product Manager of QClaw, to break down the thinking behind that frenzy, from the cultural logic to the business rationale to the product design choices shaping it all. QClaw is to be accessible to everyone on April 21. It is the first consumer-grade AI agent built on OpenClaw. No technical setup, scan a QR Code, and the agent will be live in 3 minutes. Product link: qclawsg.qq.comWaist list: https://docs.google.com/forms/d/e/1FAIpQLSeIfEzlOV8jq_tGMbV5mqTSALyufE0kZ933XqE3Fnha1_CRfA/viewform?usp=publish-editor (Founding Claw — limited 20,000 slots)Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently. Season two will host a series of guests from early-stage investing, as well as builders, founders, and product managers. For more information on the podcast series, see here.To find the previous episodes of Differentiated Understanding, see here.Chapters00:00 Introduction to OpenClaw Frenzy in China02:30 Shuyu Zhang’s Journey and Insights on AI Accessibility05:27 Cultural and Societal Factors Driving AI Adoption in China07:48 Understanding OpenClaw’s Popularity and Usage10:36 Exploring Tencent’s AI Product Ecosystem13:33 QClaw’s Integration and User Experience15:45 The Philosophy Behind QClaw’s Design18:14 Raising the Claw: The Concept of Personal AI22:48 Incremental Value in AI Products23:33 User Experience as a Priority27:55 Understanding User Needs and Safety Concerns33:07 Business Model and Global ExpansionTranscript (AI-generated)Grace Shao (00:00)Thank you so much for joining us today, Shuyu. I’m so excited to have you on the first day. I understand it’s such a pleasure to have you here. I’m really excited for today’s topic because it’s something that’s kind of been at the top of mind for a lot of people. Why was there a massive open call frenzy in China? Why did it take off? The thinking behind it all. How did WeChat go about opening up to this whole new era of agentic AI within WeChat development? And there’s no better person to talk about this topic than you. So first of all, for the audience, please start with telling us about yourself, your role. How did you end up here? Shuyu Zhang (00:35)OK.Hi, everyone. I’m Shuyu, product lead of Qclaw. Also, the architect behind it’s overnight growth story in China. I achieved breakout success with zero marketing spend. I got a master’s degree of finance from Washington University in San Luis, then worked for Alibaba Group as head of AI product in Sanyo for four years. We mostly crafted AI product for business there. Almost everything we did was about AI for work. But one day, I decided to do something different. I want to get more exposed to consumer side.to experiment with the chemistry of AI for common people. Tencent is famous for its consumer side products I want to work with and learn from consumer product experts here. So I came here last September. Yeah.Grace Shao (01:15)Really cool. And I think one thing that really kind of resonated with me is when we’re talking about AI agents right now, how a lot of times, you know, the products still don’t feel that intuitive for non-technical people. And you yourself joked and said, you know, look, you’re like a social science and liberal arts student. You’re not a technical person yourself, but you’re able to lead the product design for this. And your mission is really to make QClaw more accessible to non-technical people. Tell us about that and the thinking behind it all.Shuyu Zhang (01:43)Okay, okay. Actually, the story starts when I was working for Alibaba. Initially, we worked for the engineers to help them improve their working efficiency with AI. But later I found out that this group is overly served. They only account for like 1 % of all of the people, butEvery day there are lot of products designed exactly for them. And I think this is still the major theme of AI revolution since 2023, where people think the way to AGI lies. But these groups also are easily unsatisfied. They raise a lot of questions about the services that the AI provided, and at the same time, they’re afraid to be replaced. So I think the vibe is strange. So when I’m moving to a new environment to Shenzhen, when I was hanging around, I found a lot of interesting common people are using AI.And they’re getting a lot of happiness and convenience, even though the product capabilities for them are not the most advanced and designated for their needs. They’re still satisfied. There are two interesting stories. The first story is when I was getting home on the plane during spring festival, I met a 12 year old girl. This girl looks smart. She was playing with some toys. And after that, she was playing with a chatbot during waiting for the plane to take off. I was shocked.Because in the past, when I am younger, we usually play games during while we’re waiting for the plane to take off. But the youngest generations are playing with chatbots. This is interesting. And she’s using it to either cut the photos and even make phone calls with the chatbot. And when I talked to her, he also told me that when their friends are hanging out, the 12 years old girls are hanging out. They’re playing with the chatbot too. And I was really shocked by that story. And the second story is during the early January, Yang Liping was doing opera in Shenzhen. And I went there too. And a 50 year old lady sitting next to me, I cleared at her phone screen. The phone storage was actually out of use.But she didn’t delete the chat box applications in her phone as well. She only reserved WeChat, TikTok and the chat box applications in her phone, even though it’s already full because she might not use the very advanced phones. Yeah. And then I found out that for common people, the requirements or the needs for AI exists as well. And problems widespread for the problems, not widespread, for the problems widespread, but not complicated. But the supplies for them are far from enough. And I know there are a lot of people are chasing higher and stronger AI, but the gap between the bottom and the ceilings, 90 % of the people are between them. I want to make a product that can feel and satisfy these people’s needs. Yeah, that’s my story.Grace Shao (04:30)That’s really interesting. think it really ties into a theme that we’ve been writing a lot about AI-prone, which is also about how China is really embracing it as a full on mass market product versus I think in the West right now, AI is still really used by a select group of knowledge workers or certain kind of demographic. I do want to double click on that, which is what do you mean by the young girl is playing with AI? What was she doing actually?Was it interactive with their friends or were they trying to build products or what were they doing?Shuyu Zhang (05:00)Okay. The young girl, when she was playing with the chat bot applications, she actually sent a photo of her roughly taken, not in a very good light or in a very good background. And she just asked that, asked that application to curve it for me to make me look prettier or make me look funnier. She’s not actually a, because I asked, I also asked her a very interesting question. I asked, do you post TikTok shorts or Instagram?She said, I don’t because I don’t like to show enough my life to the public, but I just like to see my photos in the funny way. don’t even though I use AI to, you know, process my photos, I’m just enjoying it by myself. I don’t want to it to other people. And the happiness of the AI processing of the photos is already enough for her. And this is the first scenery she’s using. And the second scenery is that she actually stays at school all the time. She didn’t go home during the from Monday to Friday. So she told me when she missed her mother, because her mother worked in Shenzhen and is a working mother, her mother doesn’t have a lot of time to, you know, FaceTime with her or the teacher doesn’t allow that as well. Because when she go back to the dormitory, the roommates are silent. They can’t do that, but she can always talk to that chatbot. It’s like a companion. And that also makes me feel warm actually, but also little bit sad for her. And the third scenery for her is that she told me she would call the chatbot because the chatbot never blames her and the chatbot always holds her words because sometimes for I don’t know, for the young generations, a lot of their topics are hard to get for the friends, but She said, chatbot is always a good friend because the young generations, they don’t actually care about the or they don’t know about the appearance of people or the words behind the words. But the chatbot is always blunt and sincere and always happy to chat with. Yeah, this is the three scenarios she’s using it.Grace Shao (00:00)Following up on our previous conversation about why Chinese people seem to have a much more optimistic approach to AI and why did the open claw, why did open claw take off in China, like such like wildfire.Shuyu Zhang (00:15)Okay, so I think Chinese people in general, embrace technology with open arms. They have a strong, better self mindset. They believe that new tools can help you learn faster, work smarter and live better. People want to upgrade themselves. And the second point, and also this is a key, for over a decade, Chinese tech companies have been quietly lowering the barriers using technology. They make complex things simple. You don’t need to be an engineer to call a DD or take out on Meituan or buy anything online. It just works. So people naturally expect that new technology will make life easier, not harder. That’s exactly what Qclaw and Tencent’s lobster products did. They took OpenClaw’s powerful but geeky core and wrapped it into a simple IM plugin. The barrier to entry dropped from weeks of learning to 10 seconds. That’s why lobster caught fire in China. not because it was the most advanced AI in the world, but because someone finally made it useful for everyone.Grace Shao (01:13)I think that’s a really interesting take and I think in general it kind of ties together to the bigger kind of sentiment as well where overall the reputation of Chinese big tech such as Alibaba, dance and tents that still are perceived quite positively by the average person to be an employee there is something very prestigious, ⁓ very like sought after. Whereas in the US, I think in the last couple of years, there is a bit more contention or negativity around the big text, whether it’s monopoly or behavior or even the capital allocation that they’ve really received that’s unfair compared to the rest of the country. overall sentiment is a bit different. I think that really did contribute to this as well.Shuyu Zhang (01:58)Yeah, exactly.Grace Shao (07:00)I think that’s really eye-opening and I think I’m not trying to hijack this whole conversation, but to me when I hear that, I think I want to like you said, the companionship is really great and the ability to help the child feel more connected to her mother is great. But at the same time, I do feel like there is some concerns or worries about that. ⁓ I did notice that the Chinese regulators recently pushed out some regulations around actual child use of AI.which we’ll have an expert to join us one day to talk about this. But I think it’s interesting to showcase a phenomenon of China really embracing this at a mass scale. On that note, I don’t want to go too deep into the child use today, but on that note, I do want to ask you, why is it that China seems to be so amazed and enthused about AI, and especially this time with the open-claw embrace?Obviously at Tencent’s headquarters, saw pictures going viral where people were lining up and getting open-clawed saws. Various big tech, whether it’s Alibaba or Baidu pushed out similar products like yours, like Qclaw. Could you explain to us from the big picture, is it cultural? Is it societal reason? Is it a top-down policy reason? Is it commercialized as a business reason? Is it product design, like you said? What is it that really drove like everyone going AI.Shuyu Zhang (08:16)Okay, so the first background of Chinese AI is that since 2023, since the chat-chip goes out and after that Baidu, Alibaba and also Tencent and also most importantly DeepSeq, it’s widespread of AI, their LLM makes AI widespread in China already because during last, the one before the last Spring Festival, almost like 200 million people use DeepSeek every day. So the basic foundation of people knowing AI in China is already widespread. And why OpenClaw is also going wild in China this year? Because OpenClaw’s capability is quite different from other products that people are familiar with currently. So this is the first decision point. more about the culture thing.Firstly, China is a fast developing country and Chinese people are diligent by nature. And every generation, people of every age want to be a better self. So during this race, anxious middle-aged actually is a very big contributing factor. And also fear is also a big pusher. They always fear, you know, when they’re getting old, they will be lagging behind. So they want to, you know, learn more things.learn what the new generations or the techie guys are doing. Yeah, this is the first decision point. And the second point is that hiring a personal assistant or secretary here is not common, but everyone wants to be an emperor because there are so many operas and TV series and soap opera shows recently about how the past generations, how the ancient times, how the emperor times. Yeah, everyone wants to be an emperor. So texting a message, gets the people done your job. Everyone wants it. So as long as you make the product simple enough and convey strong similarity with being emperor by sending messages through WeChat, really suits people’s taste. Yeah, I think this is two big factors about culture and society reasons here.Grace Shao (10:18)It’s interesting because basically they’re saying involution itself needs you and has made everyone want to adopt a new technology faster. It’s something I’ve never thought about. On the second point, am curious, like jokes aside about the Emperor thing. What is it that like the average person, like what are they using OpenClaw for though?Shuyu Zhang (10:36)Okay, actually there are two big categories. The first category is still for the common people because they don’t, even though they know OpenClaw is wild, they don’t know what to use it about. So they still use it like Yuanbao or Doubao or the other phone. They just like asking, yes, yes, yes. They still use it like the chat bots. the second part, they already use it in a more advanced way. They used it to earn money.Grace Shao (10:53)which are chat bot products. Yeah.Shuyu Zhang (11:04)Like for example, ask QClaw to seek jobs for them, like scanning the boss or the Liepin website and apply for the job.Grace Shao (11:12)Which are LinkedIn, Chinese LinkedIn, Craigslist, or indeed kind of like websites. Yeah. Okay.Shuyu Zhang (11:18)Yes, yes. And the second part, they use it to operate the social media account like Red Note or they even use it to operate X account, get some posts from X and watch it to write on other social media platform. Yeah. And the third condition is that they actually, trying to use it for like investing suggestions, how to invest in some stocks, what’s the price to get in and should they keep it or sell it.Yeah, these are the main categories, but they are all about making money.Grace Shao (11:49)I see. That’s what fascinating. I want to get into case studies a bit more later. But before we get into this further, I want to help our listeners understand, can we, provide a base framework? Right now there are five claw bought like products within just Tencent ecosystem. And Qclaw is one of them, right? Which falls under WeChat, the product WeChat. Can you help us understand like these products first?Shuyu Zhang (12:13)Yeah, sure. The five claw product here is different for the users and are different between the target users. First, Qclaw and also WorkBuddy, we’re targeting at consumer and the Lighthouse, they’re targeting at enterprise side needs. And also there is a product called Claw Pro. They’re targeting at enterprise for the enterprise who want to make their stuff. Everyone has an ⁓ enterprise size claw and also the cloud desktop they’re targeting also at the enterprise side. Yeah.Grace Shao (12:44)I see that that’s just a good framework to have. So, okay, let’s get into the product pieces and your strategic intent then. OpenClaw is the open source framework ecosystem, while QClaw is Tencent’s package localized layer built on top of it. Can you actually help us understand how that works? What does it mean to have an OpenClaw integrated into a Tencent ecosystem?Shuyu Zhang (13:04)Okay, sure. Okay. So open cloud is actually a package of codes. If you install them on your laptop and connect it with LLM APIs, you can have a personal assistant already on your desktop. But that were required to handle like command lines, which is very technical skill. Even though I’m a product manager, I don’t know how to run command lines before after my engineers taught me to. Yeah. So purchasing and also purchasing APIs from providers is also not familiar for common people.and also connecting with WeChat channels or like the other channels is also not that so easy before we do it there. OK, so we made all of these coding execution into visible and simple product features, which are already educated to common people. For example, we made the channel connection by making it just scanning a QR code and you can already get a QR code onboard on WeChat.Grace Shao (13:33)Mm-hmm.Shuyu Zhang (13:59)And also we make all of the LLM API purchasing processing invisible. We don’t need them to purchase a game. We are reincarnate in the product. And also the installation part in the past, people might need to, know, NPM run open claw, but now they only need to download the applications and double click it’s on.Grace Shao (14:19)see that that’s really helpful for people who don’t understand technologies, understand how this works, including myself even, I was a bit troubled. So I want to understand the thinking behind the QClaw product, right? You kind of talked about it, the frankly, the more cultural aspect of like how this came about and your own personal mission. What was the business decision really for WeChat? Like why did WeChat push out QClaw?Shuyu Zhang (14:43)Okay, so the thinking behind QClaw, how did that come out? Actually, aside from the thinking that I want to build a product that is easy enough for common people to use it, I still have the following thinking. The first is, ⁓ what’s the vibe of the product we should use? Is it work or life or both? Because the chosen, the choice of the vibe will be different for different people. How do we categorize that the work it can do for us? Because in the past, I think all of the AI products categorization are hard to get because most of time you just categorize, for example, something like finance or.⁓ work usage or something like information gathering. my God, who knows that? So, and also there are a lot of times that work and life are mingled together. So if you’re designing an agent or a product features for everyone, if you’re trying to do that, that’d be hard. For example, if I want to build a finance agent, common people might just want to, you know, like search the stock price of something for me.or recommend whether I buy or sell. But if for the professional users, their requirements will be higher. For example, they want, they wanted to, for example, write a quantitative trading strategy for me or something like that. Actually, the depth of capabilities providing would be hard to define here. And also it may discourage users if not handled properly because ror example, if for the pro users you’re designing it too easy, they would think, this is useless. This is far from replacing my interns or something like that. And if it’s designed to be too complicated, the common users would think, my God, this problem is not just designed for me. I don’t deserve to use that. And people are born to have fearness towards their unfamiliar domains. So I think technology or products shouldundermine these fears or lower the barriers here. So in that way, we actually divide the categories of QClaw in three ways, which will be in international version. We categorize in three ways. First, QClaw it up. For the things you don’t want to do, but I have to let QClaw do it for me. That’s QClaw it. And also,QClaw daily for the things I need to do every day, but I don’t want to forget a break. And the third QClaw up for the things I can achieve by myself and the expertise support. We divide the categories in these three ways. So everyone would have the it daily and up requirements. And we will also be more flexible or more concentrated focus on what types and what level of capabilities we’re providing.This is the first thinking behind that. Yeah, because I think that the categories are complex and I want to make it simple and direct and focused to the people’s needs. They will know this, oh, this is for me. This is not too hard or something I don’t deserve. This is something I deserve to use and it would really help me. And the second thoughts behind that is, it’s a line we draw in selecting the building clause for agent.since we already raised a lot of full grown claw. Because in China, a lot of people find it hard to raise a claw. They have to educate it. They have to do a lot of configs. So it’s hard for them to raise. But we...Grace Shao (18:05)Sorry, one moment. Explain thecontext of what raising a claw mean. Like in Chinese right now, the buzzword is 養龍蝦. Explain to people what that means.Shuyu Zhang (18:14)Yes. So raising the club, a lot of people for common people, they were thinking like feeding all of my knowledge, what I know, who I am, what I want, what I like to it. They take into your input and they know what to generate in this mind. And also there is a mechanism called dream during the dream. They were, they were rethink about everything you, you told it today and they would generate something called memory.And in the later usage, they will use this memory to know you better. So you will know that after daily’s inputs, after daily’s talk with it, your claw will know you better. And every instructions you give it will be better than the common AI products that don’t know you. That is called Yang Longxia or raising the claw.Grace Shao (18:59)It’s so funny. It’sbasically providing the technology, the context, but then when it gets better, better people say they’re like lobsters are growing, growing. It’s a funny analogy. I don’t know how it caught up, but it’s hilarious. Yeah. But yes, please continue. Thank you for that context.Shuyu Zhang (19:12)Yeah, sure.Yeah, yeah. And I also think Yang Longxia or raising the claw is interesting because it’s like raising a kid or raising a pet. Because in the past, I think everyone still remembers there is a product called QQ Pet on QQ of Tencent. And I think back in my days, I was like seven to nine years old. I also raised a pet by myself, even though it died twice. Yeah.Yeah, I really enjoyed raising a pet, like feeding it every day or just bring it up on the website and see it on my desktop every day. I think that is interesting for me. And I think a lot of people would want that this kind of companion of AI and they also enjoy the feeling that something is getting smarter or clever because of them. There is connections between them and the AI. They will bring a great joyness here.not just something, for example, something that is already very supreme, high end package well and bring it to you and you just use it. You don’t feel connections with it. I think this is a very different feelings here. OK, so keep going with my point. ⁓Grace Shao (20:22)No, I think it’s funnybecause I do think you touch on something basically like how Chinese tech companies gamified as well. So that’s how it also helped the mass market adoption that we were talking about earlier. And you just reminded me like when I was young, we all had Neil pets. think any millennial people in the West would know that. And like, even though they’re a virtual pet, you actually had a strong emotional connection with it. So I kind of see what you mean by this whereA lot of people might have not even found a purpose or use case originally, but even just building that context, that relationship with the AI that actually helped, you know, adoption rate. then sooner or later you try to find ways to make it more useful. Right. But yeah, please continue.Shuyu Zhang (21:03)Yes.Yeah, exactly. OK, so after I saw that trend and also I saw the problem here, I think if we’re giving them some ground-claw, we should bring incremental value or incremental user experience to these people. And because this is important for the product perception, because if we’re the same like the chatbot about what we can do, people wouldn’t think or people wouldn’t take it seriously.people will still think, okay, this is just another chatbot, but I want to bring incremental value to them. For example, before, in the past, when people want to make some travel plans using AI chatbot, they can only say like, I’m going to like Shenzhen for three days trip. Can you design a trip for me? Because in the past, like even though AI might be different in every answer, there are 80 % of the answers are similar.or told you like to go to some park or some supermarket or some something like that or the hotels to say but if you’re using claw it would told you okay based on your fondness based on your habits based on the things you told me before i think some blah blah blah hotels would be better for you and some restaurant which is closer for example to your hometown or the flavor is similar to what you have told me that you likeThis would be the incremental value. also the claw can also like book the flights, book the hotels, or just, you know, make transactions with the restaurants and ask for example, I’m going to spend my birthday there. Can you arrange something for me? All of the things is incremental value of QClaw AI can bring to people compared with the common air products. And I think this is the second thought behind it.I want to do something that brings incremental value. Yeah. And I think these are the considerations here.Grace Shao (22:48)So I think one thing that’s quite interesting is Qclaw is obviously built out by the Tencent team. However, it can be accessed not only through your own WeChat Wecom, which is the WeChat Enterprise product. It can also be accessed through ByteDance Lark, which is like the Slack product within ByteDance. What is the thinking behind that? Why did you open up to your competitors essentially?Shuyu Zhang (23:06)Okay, so these are all channels, channels where people are already living in. We want QQL to keep company with users either in life or work or any interface. So limiting any channel will bring inconvenience to users. For us, the user experience is ultimate mode. So we don’t really care about the other, you know, so-called the business consideration. I think the user experience is the most important thing for us. Yeah.Grace Shao (23:33)That’s a very WeChat answer. feel like Alan Zhang has been kind of known to always prioritize user experience over any other kind of thinking, whether it’s commercialization or even, you know, sometimes functional adjacency within other products within the Tense Umbrella. So interesting. OK, so I have more questions. Something we talked about prior to recording was that you said WeChat is the default entry point for QClaw.And that’s still a huge mode or advantage for you guys, especially certain features that you guys introduced, such as like scanning the QR code for downloading or installing a claw has been a big selling point. Walk us through kind of why is that and why WeChat, QCOP being built within WeChat is something so powerful.Shuyu Zhang (24:22)Okay, so actually there is a very simple reason before that and after that I will explain a more complicated one. The very first simple reason is that during the initial launch, we only had five engineers and me in the very first place. And we started to develop this product after spring festival, but we launched it in March 9th, I remember. So the time is very limited, but there are so many things to be productized.of OpenClaw. What is the choice here? Because I want to make it simple. I want to make it user friendly. I want to make it widespread. So the first two things need to be adjusted by us. We need to do deletions. And the third part, we need to do adding items. Because before that, actually WeChat is not supported by OpenClaw officially. So we justwatch all of the files on the WeChat open platform and found out that there is actually a way to connect WeChat to OpenClaw. Then we just implemented it. And actually we don’t have more time to do that. So I think, okay, if I only have one time to make a decision or I only have one chance to select the channels, what would I choose? Of course I would choose WeChat because not everyone use some other working messaging applications.But almost everyone in China use WeChat, even though whether you are like four or five years old even, or you’re 70, 80 years old, everyone use that. so that is why I choose it for the first channels. And the second reason is that people are mostly adjusted to use WeChat. And WeChat can be also connected through scanning QR code, which is the user experience other products can give them.because other products, you see it on the light, like the tutorial, you need to go to the open platform of that product. need to copy your user ID, which is a very long link, and you need to copy the token or pin or something that is so technical terms. People would get scared by that. But we already have a very simple user experience, ⁓ user interface method that is getting QR code.Why don’t we just do that? And also, also through the past years, mobile payment through scanning QR code is also widespread by Tencent. Tencent made scanning QR code and making payments widespread in China. And everyone, either they’re like the merchandising, the business, big business or small business, they can have their own QR code. And everyone isused to scanning the QR code and connect everything. That is why we choose WeChat as the default selection.Grace Shao (27:04)I think that’s interesting because it’s like basically what headlines been missing. A lot of it is also just the native user experience and the ease of people to even access this kind of new technology because I think like you said, a lot of people are not maybe not scared, but it’s intimidating, right? It’s intimidating to try out new things and a lot of what’s out there in the market right now feels very technical. And if you’re not a technical person, you’re not following the progression of AI like closely. It’sIt feels intimidating to even try these new products out. Yeah, so I want to bring the conversation back to the real life usage, right? Because that’s a thread we kind of been talking about throughout our conversation. You really focus on bringing AI agents to the average show, the average person to the mass market. What is it that people really, really want out of this? Is it?purely for gimmicky use, it’s for fun. You know, the case that he brought up with the young girl is obviously very interesting, but I would assume that’s not the mass demographic, right? Is it for consumer convenience, prosumer productivity? Is it for small medium sized businesses automation? I wanted to understand that. And then I want to expand beyond that, which is, are they not concerned at all about safety or privacy when they’re using these products?Shuyu Zhang (28:22)⁓ actually I think, yeah. Okay. So, ⁓ the, the first question I’ll explain here is what is a real target use case here? Either it’s consumer convenience, prosumer productivity or small business office of my automation there, right?Grace Shao (28:23)So two parts.Shuyu Zhang (28:36)Okay, so for me, it’s still consumer convenience, or we don’t categorize in this way, because we look at people as the subject, what the people need to do and how we can make it smooth and convenient. People can have different requirements in different conditions, either it’s on life or it’s on productivity, or some more serious or related to the business. We make integrations and push the ecosystem to provide the rest.We will, like for example, if you see QClaw international version, you will see already put like some, I mentioned before the QClaw and QClaw Daily about for example, either you’re seeking jobs or you are operating your ex account or you are, for example, your career pop fan and you want to search for the concert tickets or chasing all of the information behind the hero. This is the conditions we provide. And we will also push the ecosystem.For example, we already have a lot of ecosystem supporters here who provide the doc intelligence, like providing, for example, scanning contracts, the receipts, scanning something like that. And also there are ecosystem friends who already provide video generation, something relevant to, for example, creative parts or the designer shop. Yeah, we have a lot of ecosystem friends.providing here. So we are still starting from what the people needs in aggregate, what people needs aggregately. so this will, the second question about the safety problem here is that we actually provide an incarnate safety features called the AI gateway or in Chinese Longxia Guanjia. That is something relevant to our team because my team is a Tencent PC manager.which is a very, very, very old product. I think many people who knows this product might be like 30 or 40 or even older. Yeah, this product was built in 2004. remember, might not be correct. So it’s still in the computer-sized format, but in the past, it actually used a lot of safety capabilities.Either it’s like preventing prompt injection or skill security or a lot of file security. It already has a lot of capabilities in it. And it’s also vetting the possible cyber attack over the internet. So we will know what’s the risks here. So we already provide a gateway in the product that everyone can be protected under this gateway.they will not be exposed to the risks on the internet already. And also there’s point that, yeah, and actually.Grace Shao (31:18)I see. that actually protectsthem more than someone directly installing an open claw themselves, right? Like going through cue claw is a lot safer in that sense.Shuyu Zhang (31:28)pardon?Grace Shao (31:29)So it’s basically a lot safer for the average person to use Qclaw than installing their own open claw on a Mac Mini or whatever. Because there’s not that kind of safety guard rail built around it, is what I’m trying to say, yeah.Shuyu Zhang (31:37)Yes, yes.Yes, yes, because common people actually initially if they’re as long as they’re on the internet, they’re exposed to these risks. That is worse when they’re using open cloth without any protection. ButGrace Shao (31:49)Mm-hmm. Yes.Shuyu Zhang (31:55)There’s also interesting part is your computer actually don’t have a lot of information and you don’t have a lot of money. So you’re not actually a target. Yeah. But we provide enough to safeguard for these people. And for the more important issues or more severe issues, part of this will provide higher levels of security. Yeah.Grace Shao (32:04)Yeah.I see, Yeah, I think, you know, that’s a really good big picture on just QClaw’s build out and why China really took on like QClaw at a mass scale. I want to understand your business model and long-term implications. Obviously, I understand, you know, you guys fall under Tencent. It’s extremely lucrative business in the cloud side, the gaming side, obviously, we chat advertisement, etc.You guys might not face the pressure to make money from this product, but I still want to understand how does it work? Will QClaw be free basically forever? Will you guys introduce a subscription model? know, I know recently you’ve even talked about you’ve been traveling around the world a lot. You’re in the States. You’ve been in Europe for some time. Are you trying to go global with this product? Are you trying to sell globally? Is it to enterprise? What is the kind of business behind thinking behind this?Shuyu Zhang (33:08)Okay, so the firstly, it will not be free forever, but we will always provide some free tokens for the first time users because they deserve to know what it can bring you, what increment value it can bring you. So we will provide some free tokens here, but we also provide different subscription plan here to support different layers of requirements. But I don’t think the token fee will be the only monetization methods here if we bring people’s whole life here, just like the...WeChat strategy as well. Yeah, because selling tokens, tokens currently is really expensive. Even it’s a huge company, will still face the pressure here. Yeah, but a lot other modernization methods here, but we will also be very cautious here to trading between the experience and the modernization here. And this is theGrace Shao (33:46)Mm-hmm.Shuyu Zhang (33:57)the answer about the subscription plan and the charging plan. And also we are traveling abroad. We are trying to go in global. And this month, exactly April, we’ll be launching internationally and we will be starting from some main regions, North America, of course, and Asia Pacific. And then we were spending into more areas. Why? Because I traveled a lot in the last years. Everywhere I went, I would stay there for like a month.And I would thoroughly experience the real life there and talk to the people there. I found out that even though there are differences between people’s life, of course, but people everywhere share similarities in needs, and they also have curiosity towards others. So people are bringing a better way to live with themselves with QQLO, like QQLO A, QQLO WAP, and QQLO Daily. So they can share it with others.on Qthaw and benefit from other people’s sharing. That is why we are bringing it globally. And every states we go, we will co-create with people there and pass on merit to more areas.Grace Shao (35:00)Yeah, so I think it’s really interesting because obviously QQLA has a huge advantage in China because it leans into the WeChat ecosystem we talked about a lot today. But what is your advantage when you are going global? How do you compete with international competitors?Shuyu Zhang (35:16)Okay. I think the first important part is still the use experience because I use a lot of global applications because I, my job is AI product manager. I would find a lot of product even complicated for me. I think the product experience is not, hasn’t been done very good yet. There are still a lot of bugs. There are still a lot of, you know, complex, complex items, complex terms or complex workflows here.I want to make it all simple. And integration would also be a great part here. And the third part is the community advantage. would, you know, because we are not so intimidating in the image, we will not be like, we are the tech guys. We are the advanced ones. If you don’t know it, you’re the stupid one. You’re going to learn from us. No, we will not do it like that. Yeah. Yeah. We would co-work with the creators. For example,Grace Shao (36:05)YouShuyu Zhang (36:11)⁓ For example, when we’re spending to Japan, will co-create with some local, like fan, big fans there, or the ones who knows well about the food there, or the one who knows about the job marketing there. And we will make it more localized and make the people who really use it create that. We will co-create with them and bring it.to their community because that would be the way that the community gets the concept or gets the usage of AI in fastest way. Yeah, I think so. basically.Grace Shao (36:44)Yeah, I see what you mean. Like you marketed a much more accessible product than other peers maybe on the market right now, which are targeted for again, relatively niche demographic. So that brings me to the next question, which is like, do you think then the messaging apps with a chatbot like kind of interface will become, will remain the default or will we see a new kind of interface layer for agentic AI and how we call on them.Shuyu Zhang (37:16)think, actually, I don’t know about this answer because messaging apps can be, they can evolve as well because they have a lot of engineers as well and they have a lot of ⁓ intelligent people there and they care about the user experience here and Asian products can also evolve. The true interface layer, I think is dynamic and there is no fall or lamb in the current arena. So I actually don’t know the real answer of the, you know, who are the final interface layer, who is the true one? I don’t have the answer yet because I see a lot of interesting apps evolving from AI agents, but they’re trying to, you know, cutting into the messaging apps interface. And also there are a lot of messaging apps there. for someone like the X or I don’t know what’s his future plan, but I also know there are a lot of messaging apps who are cutting to an agent domain. the answer is dynamic, I think.Grace Shao (38:13)I appreciate the honesty and the humility actually. I think no one really knows the future right now. The speed of evolution of industry is insane right now. And I always hear people who really like, you know, like even the Ben Evans and the world, they’re like saying, if you think you really know the industry, you don’t really know the industry because you can’t possibly, you know, have a strong grip on what’s happening because things are changing so fast and so much happening constantly.Grace Shao (39:05)I have the last question, which is a question I ask every single guest that come on the pod. What is one differentiative view you hold? This can be anything about the product we talked about today. It could be about the industry. It could be about anything in life.Shuyu Zhang (39:08)Yeah.Okay, one differentiated view I hold is that I think the most profound function of a superior AI, like the clock, is not to solve problems, but to reveal them, to reflect back to us the questions we’ve been unwilling to ask ourselves. Think of it as an archeologist of behavior. We narrate our lives, edit our memories, even lie to ourselves without knowing. But something like the clock observeswhat we actually do, what we choose or what we linger on. It doesn’t judge. It mirrors. In the end, what it shows us isn’t its own intelligence. It’s us, our contradictions, our desires, the fractures in our collective consciousness. So the most meaningful conversation isn’t whether AI is becoming human, but whether we are brave enough to look clearly into this vast, mirror it holds up. And finally, see...ourselves.Grace Shao (40:18)I think that’s super interesting think it helps us. You touch on something, it’s like really helping us recognize blind spots. I just want to share a personal anecdote as well, which is like, think, you know, when I started AI Pro nearly two years ago, it was really hard for me to sometimes seek help from other people’s opinion because it takes people’s time, right? And then for reviewing of my work and I didn’t simply want a grammar.like, you know, copy editor kind of grammar fix. So now what I did was I built an editor council and what I did is train the council basically to have certain perspectives, follow certain guidelines or, you know, ⁓ angles of the world and then critique my work and really help me recognize ⁓ blind spots I’ve been missing or my logic or my thinking that are not, you know, synthesized clearly or not.flowing smoothly, things like that, that I just found it so helpful. In fact, in some ways more helpful than a human editor at certain tasks, because exactly to your point, humans have biases, humans have judgment, and not like intentionally, but just by nature, we all hold biases based on our own knowledge, whatever. But the machines basically are just can be very critical if you tell it to be critical and can kind of show you a 360 view of your thinking. that’s super interesting and I appreciate your sharing on that.Okay, thank you so much for your time today, Shuyu. If anyone wants to reach out to you, how should they find you? If they want to learn more about QClaw, where should they go?Shuyu Zhang (41:50)my LinkedIn. The name is Shu Yuzhang. And they can also follow me on my X account, which I can share with you later. It’s also called Shu Yuzhang. Okay.Grace Shao (41:59)Perfect. Thank you so much. We wonderful conversation with you. Thanks again.Shuyu Zhang (42:03)Thank you, thank you Grace. Bye.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Sovereign AI, Open Source, and the Gulf’s Big Bet with Interconnected Kevin Xu 14.01.2026 58λ

  Every panel on AI and geopolitics seems to default to the same cliché: “the US–China race.” In this episode of Differential Understanding, I wanted to sit with someone who has actually lived inside DC, Silicon Valley, and the US–China tech corridor, and ask whether that framing still makes sense.My guest is Kevin Xu, founder of Interconnected Capital – a global hedge fund focused on the picks and shovels of AI – and author of the Interconnected newsletter, which sits at the intersection of tech, business, and geopolitics. Kevin’s path runs from Obama campaign staffer and White House / Commerce Department comms to GitHub’s international expansion lead, and now to full-time investor–writer with a very explicit geopolitical lens.We start with why he insists on “thinking in public” as an investor, and why he believes ideas soulocking in a vault. From there, we dive into his critique of the “race” narrative and his alternative concept of US–China co-opetition – a messy mix of competition, cooperation, and outright co-opting of each other’s models and research. That leads naturally into China’s open-source AI ecosystem, the Manus–Meta deal, and what he would need to see before feeling comfortable owning the upcoming MiniMax and Zhipu IPOs in Hong Kong.In the second half, we zoom out to sovereign AI: why South Korea might be one of the few countries outside the US and China with a shot at true full-stack AI sovereignty; how to read OpenAI’s Stargate initiative as an explicit American export play; and why the Gulf – particularly the UAE – is emerging as an AI “swing vote”, combining abundant energy, sovereign wealth, and a 1.5 million-strong construction workforce into a potential global compute hub. We close with Kevin’s differentiated view on China: AI diffusion is far more visible there, but the economic impact is not necessarily greater, and Beijing may end up being the first government forced to confront AI’s social implications.In today’s world, there’s no shortage of information. Knowledge is abundant, perspectives are everywhere. But true insight doesn’t come from access alone—it comes from differentiated understanding. It’s the ability to piece together scattered signals, cut through the noise and clutter, and form a clear, original perspective on a situation, a trend, a business, or a person. That’s what makes understanding powerful.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.For more information on the podcast series, see here.Timestamps (chapters):* 00:57 – From DC to GitHub to Interconnected Capital* 02:39 – Why Kevin “thinks in public” and writes Interconnected* 04:44 – US–China AI is not a “race”: co-opetition explained* 09:27 – Open-source / open-weight AI as last bastion of global cooperation* 12:14 – Capital flows, decoupling and why “capital finds a way”* 15:36 – Manus x Meta: product quality, viral growth, rationality in AI* 19:40 – MiniMax & Zhipu IPOs: revenue reality vs AI lab hype* 24:39 – Can Chinese labs win the Global South with cheaper AI?* 27:09 – Sovereign AI 101 and why South Korea looks uniquely powerful* 32:03 – Stargate as de-facto US sovereign AI and export strategy* 34:44 – Kevin’s trip to UAE: Gulf AI strategies and the “swing vote” thesis* 40:06 – Sovereign funds, MGX, and attracting talent from Hong Kong & beyond* 44:18 – Non-consensus bet: UAE Stargate as a global compute hub* 46:57 – Differentiated views on China AI diffusion and economic impact* 51:29 – Embodied AI, “aunties pressing elevator buttons” and social risk* 55:01 – Robotaxis, delivery drivers, and why China may go slower than expectedAI-generated transcriptGrace Shao (00:00)Hey everyone, welcome back to another episode of Differential Understanding. This is your host, Grace Shao And joining me today is Kevin XuKevin Xu is the founder of Interconnected Capital, a global hedge fund focused on the picks and shovels of AI. He writes the Interconnected newsletter on SubSac, which covers tech, business, and geopolitics.His insights have been frequently cited by the New York Times, Bloomberg, Economist, CNBC Information, Financial Times, Wall Street Journal, among many other media outlets. He previously worked as a senior executive at GitHub, the world’s largest developer platform, and served in the White House and Commerce Department during the Obama administration. He studied international relations at Brown University and law and computer science at Stanford.Grace Shao (00:40) Hey Kevin, thank you so much for joining us today. I already introduced you, but for listeners who may not know you that well, can you introduce yourself, the different hats you wear today, running Interconnected Capital, writing Interconnected Newsletter, and operating at the intersection of US, Asia, tech, geopolitics, and investing.Kevin Xu (00:57) Yeah, first of all, thank you for having me. So as you mentioned, what I do currently during the day is I write the interconnected newsletter on the intersections of geopolitics, technology, and business. I also run my own long only fund called Interconnected Capital, focused on the picks and shovels of AI, both hardware and software. Prior to that, I actually work as an operator inside multiple Silicon Valley tech startups. The most recent one is GitHub, which is the Microsoft-owned developerplatform. I was their lead for international expansion strategy. That was my most recent real job, if you will. I also spent a bunch of time at different startups of varying sizes. And before that, actually started my career in politics. So I joined the first Obama administration’s campaign back in 2008. I was a campaign staffer. That was my first job out of college and then moved with the campaign team to DC, ⁓ worked in a few different roles.in the Commerce Department as well as the White House doing mostly press and communications work. So that is ⁓ my sort of all over the place background that led me to what I’m doing today, which is investing and writing ⁓ in technology, but with a very heavy geopolitical lens to the process.Grace Shao (02:13) I think that’s really interesting and explains to why you have a geopolitical lens, given that you actually have a DC background, right? But you run a fund and you actually keep most of your thinking, public, So instead of just keeping it mostly all private, which is what most investors do, why do you publish a very opinionated, very insightful sub stack and, share it very, I think generously with the public?What kind of conversation gap are you trying to fill when you start the interconnected newsletter?Kevin Xu (02:39) I think there are two elements to that. First is that this is more of my Silicon Valley ethos, which is that no idea is that worth keeping in secret. It’s all about the execution. Like I’m not, I didn’t come from a Wall Street finance background, right? Where a proprietary trading algorithm or some secret information you got from a meeting is this big trade secret that you want to lock into a vault inside Goldman Sachs or whatever. And that’s going to make you billions of dollars. That is not my approach to investing.I think thinking in public, sharing in public, and really getting the feedback that I get from writing is much more valuable than keeping all these thoughts in my head as if they’re the next best thing since sliced bread. When you actually write it down, when you put it out into the internet, half of them are good, half of them are actually crap. And I use writing and writing in public specifically basically as a canvas for meto think better, to hold my thoughts more clearly. I think knowing how to think is the most important skill for any investor to be able to succeed for the long term. And if any idea that I shared out there benefits somebody else, and you made some money off of it for free, so be it. Good for you that you actually understood some of the value from the writing, even perhaps more than I did as the writer. But for me, that’s not something that I keep very possessively as a trade secret.Grace Shao (04:00) I really relate to that and I think exactly to your point you’re like writing everything down is the way of thinking through your thoughts sometimes it’s all jumbled up in there and then also people ask me why do you keep AI Pro all free? I was like well if it really benefits you you know I don’t really mind I’m not trying to make money off of like selling you know just my content but really the content is my thinking and to your point sometimes I put in so much work and then the result and feedback is so bad and then some things I just kind of like throw out thereAnd then it actually really sticks with people you never know. It’s really good to get the feedback from the public as well. Well, I think now I want to ask you about your journey into investing and really covering China and US. So you are based in the US, but you are Chinese.birth, right? ⁓ So does that play into why you cover China-U.S. related work right now? And I do want to talk about your recent article, which you said you think calling the China-U.S. relationship in tech and AI a race is quite lazy. You said instead of seeing it as pure competition, you think it’s more of a competition. So cooperation plus competition. Do walk us through that and how you kind of came out with this frame.Kevin Xu (04:44) Correct.So just to put a finer print on it, as far as the personal history is concerned, I was born in China. I moved to Canada when I was little, similar to you, think, Grace. And I moved to US later on. So obviously, I work in the US government. So I’m a US citizen. So I’m actually a card-carrying Canadian as well as an American right now. And I think having had a very global citizen-ishGrace Shao (05:22) Mm-hmm.Kevin Xu (05:30) upbringing and life experience was the lens that wanted to bring to my newsletter when I first started writing it roughly five and a half, six years ago. Some of it has to do with the US China. Some of it actually just has to do with a specific industry trend in the software, in hardware, in the nerdy techie stuff, in open source that I like to talk about. And I think only the US China stuff got picked up for one reason or another. People started to pay more attention toGrace Shao (05:31) Mm-hmm.Kevin Xu (05:58) tothe stuff they’re writing when there is a China-US lens. And maybe it’s just because there’s a dearth of content out there that actually brings a level of nuance to the conversation. And that brings to what you asked me about, which is this notion of a US-China AI co-opetition, is how I like to call it, as opposed to calling it a race, which is the kind of intellectually lazy approach that I have fallen into multiple times.Throughout my own writing just calling the race calling it a race But not really thinking what that implies which is that one a race implies that there is an end point There is a finish line to this race But for AI there really isn’tEven the most fervent believer of what an AGI is does not believe that is a static endpoint in which once you reach it, you’re done. And of course, there is the implication that this whole thing is a very zero sum dynamic if you call it a race. But in reality, if you look at everything that’s actually happening on the ground between the US and China on AI, it is a manifestation of co-op petition, which is that there’s a lot of competition between different firms.between different labs, both within China and between the US and China, lots of startups. There is also a lot of cooperation. The cooperation stuff gets probably shoved over to the side or doesn’t get mentioned as much because of the geopolitical toxicity of the conversation. But there are lots of papers, academic institutions, adjoined productions in terms of research and collaboration that is still happening both in academia and frankly in a lot of startups where the approach to all this is much more pragmaticandless geopolitical. And then the last element I actually want to introduce to this fake word is co-opting. There’s a lot of co-opting between leading AI labs from both sides. When the initial chat GBT moment happened three years ago, every single Chinese lab more or less used Lama as their basic building block.to advance their ⁓ model building. Every single hyperseal in China used Lama as one of their leading cloud services to get things going, right? That is a co-opting of an American, I guess, production, if you will, of a model, just to use model as an example. And then as Chinese open source became much more well-known, much more prevalent, much more popular from B-seq to Quinn to whatever, now we have Airbnb being one of the biggest users of Quinn.We have a UiPath being one of the biggest users of Quinn and a bunch of startups that they don’t want to talk about using Chinese models to really bring down their own costs so they can run a profitable startup, co-opting each other’s work. So I think co-op petition is the most accurate way to talk about it, but I also understand it’s probably not the easiest way to say the word. And so I’m not...counting on the word catching on at all, but at least for my own intellectual honesty sake, that is the word or the way that I plan to talk about this dynamic going forward because I think it’s the most accurate way to reflect reality on theGrace Shao (08:56) I think definitely your writing is one of the more nuanced kind of work that I’ve come across on the internet where it does touch on China, US, where it talks about the cooperation as well as competition and give the audience a geopolitical background ⁓ but still focus on the business, the society and offer that more neutral un biased, I think, analysis of the businesses,But from where you sit, where do you think the founders, engineers, investors actually feel like they are really collaborating? Give me some more concrete examples.Kevin Xu (09:27) I think open source AI, open weight AI, the rise of that is probably the best and most concrete example of collaboration and cooperation happening despite all the resistance, the challenges to cooperating, right? There is a lot of resistance to cooperating on anything. And the natural way is to kind of go towards the path of least resistance. But something that is happening that is, I think, probably the biggestin 2025 is the rise of China’s open AI ecosystem becoming all of a sudden leading the world. Not just pretty good, not just, oh, it’s also happening, but is flooding the zone as far as models are concerned. And the nature of open source is open collaboration. There is no deep-seek open model.without the lineage of all the innovation that came out of GPT-2 that was actually open source back in the days, or Lama, or whatever the open things that the US lab...was feeling comfortable doing until it no longer felt comfortable doing. And then you need a lot of Chinese labs to give back to the whole ecosystem as well, entirely without charge. That is the other thing about open source is that you can do whatever you want with open source product for the most part. And DeepSeek and Quinn really led the way from not just opening it, but also having the legal license to permitjust proliferation everywhere. You can do anything with a Quinn model. You don’t have to tell Alibaba you’re doing something. You don’t have to really pay Alibaba a cent. You don’t have to even give credit to the Alibaba team. Just kind of go forth and prosper, right? Now it’s very hard to track.⁓ what that diffusion really looks like. Having worked at GitHub, for example, which is the home of all open source code for the entire internet pretty much prior to AI, I know how hard it is to track. We’ve tried to do that internally with our data. We have some rough sense of which country is contributing on GitHub more than other country, which company, but we don’t ever get too deep into the people behind that for privacy reasons and whatnot.But you know from a institution perspective and an intuitive sense that it’s gonna proliferate everywhere, right? And the only surprising thing is that this came out of China, which shocked a lot of people. I don’t know why it should shock a lot of people, but it did. But.But that’s kind of where the big story comes from. So I think cooperation is happening regardless. And open source is probably sort of this last bastion of global collaboration as the world splinters into its own camp as geopolitics and AI kind of co-mingle together to make everything feel more cagey. This is still the last kind of remaining source of cooperation.Grace Shao (12:14) I think you talk about the technology being much more cooperative than people expect or want to admit. But what about capital? Over the years, we’ve seen that. first for context, think people need to understand in the 90s and early 2000s, US capital were the predominant capital that were actually behind a lot of the Chinese big tech we see today. But today, now we know there is a decoupling in terms of US investment into China, especially in the sensitive areas such as AI, robotics, andsemiconductors, right? So do you think this is something structural or cyclical? Like, are we going to see more opening up from the US government to allow these US funds to invest in China again? Because a lot of them are obviously still interested in doing so.Kevin Xu (12:57) I think the rumor is it is loosening up. I think there’s a lot of chatter that Chinese VCs who for a period of time just could not raise any USD fundfor probably like five to six years or so is starting to do so again and I think that spigot is slowly but surely going to open up and it probably won’t be like as wide open as it used to be before but my personal feeling is that capital finds a way it’s just like water it’s going to flow towards whatever the final destination it needs to go to even if it has to go around mountains it has to go through a bunch of rocks it’s going to grind that rock to a smooth edgethat being geopolitics sooner or later, but it will probably take more time than most people have the patience for. And you know, to come back to what I talked about open source real quick, if we can double click on that, I think the contrast between capital flow and source code flow in terms of open source is that ⁓ engineers, doesn’t matter which country you come from, want to work on⁓ the most open piece of software or code that is open source and you can collaborate with the rest of world, right? Like that’s why GitHub became so popular because engineers, whether you’re from China or the US or Germany, you identify with the code that you build. You don’t necessarily identify as much with the nationality that you were born into. That isn’t really a big part of your work at all.Right? Even calling something a Chinese open model is a bit of an anathema because like what is it? What part does it really is Chinese versus when it’s out in the open, it’s just like this piece of common good in the internet now. Right? Like no one can really control it. So what’s the point of calling a Chinese or American or whatever? And that’s how engineers like to operate.So that’s why there’s this tug and pull between the geopolitics force and really the engineering and the builder force that is by definition very global.Grace Shao (14:51) Yeah, I think the engineers and scientists you speak to definitely are not geopolitically driven or as ideology driven as I think sometimes the business people because they need the support of their government for certain policies. So I think the business people who seem to sound geopolitically driven are not actually geopolitically driven. They just need to do so before for their business survival. And that’s just the reality of how the businesses work. Right. So I want to put you on the spot. We touch on this quickly before we start recording.Manus, speaking of the most famous US injection into a Chinese AI company is Manus. And I just woke up to the news, ⁓ day of recording is December 30th, that, you know, Manus was just bought out by MetaI I used to manage this really good product. How do you view this whole thing?Kevin Xu (15:36) I also use Manus. I think I got an early access code actually before it even launched. I was going to say back in the days, but that was only like a few months ago. It was actually like less than a year ago, right? And this company ⁓ went from zero to a hundred million dollar ARR in about eight months, which is just astronomical.Grace Shao (15:40) Yeah.It’s so crazy. Yeah.Kevin Xu (15:57) ⁓ growth on the back of essentially its product quality. And I think that is one of the most interesting takeaway for me as an investor, as a technologist, which is that we talk all about like geopolitics and, you know, this and that none of this is actually about the product or the tool, right? About AI, but Manas, this little bitty startup, basically proved all of us wrong, which is that ⁓ product quality still matters.if you have goodQuality product people will share you people will talk about you people will you know? Do word-of-mouth to tell other people to use your product I think one of their more famous element is their ability to kind of crack this black magic of viral marketing without spending any money Right back in the days when they first shared their first version, you know, Jack Dorsey tweeted about it all these like Silicon Valley Luminary started sharing about it and it’s because their product actually spokefor itself. And it continued to evolve very, very quickly to capture not just attention, but actually revenue, which is very, very hard in this current climate of AI kind of bubble-licious noisiness that we are living through. And on the outcome itself, first of all, congratulations to the entire team. I think it’s very impressive, this outcome to be bought out by Meta. At this moment, we don’t know how much it actually paid for. Maybe by the time this app was released, we actually know how much Metapaid for, but the last round they raised that was $500 million valuation, right? Which in AI land is actually really, really cheap because we have, you know, 10 to $20 billion startups being funded in the U.S. right now that has zero product, zero revenue, and more or less a bet on a very impressive team, which could still come out okay, but we will see what happens. But I think this Manus dealto me is a very I want to say it’s evidence that rationality still matters. It’s evidence that like economic kind of pragmatism still has its moment in the day and doesn’t have to be whiplashed by geopolitical consideration. So I find that the deal very heartwarming as an investor who really just hopes for more economic rationality for everybody who’s involved.Grace Shao (18:18) Yeah, I think ⁓ to your point, like it’s definitely like a positive signal because it means that people are evaluating the products how good they are instead of just the narrative of the geopolitical kind of cloud above it. And I think it’s really interesting. Like when I was speaking to people like about this deal this morning, they’re saying actually, you know, people overestimate the PR they done back in the day when they first released it. It wasn’t because, you know, they did some black magic PR.It was simply because they didn’t even have the compute capability to actually serve too many people. So they sent it to people to try first. And I think I have a lot of startups that come to me being like, how do I achieve madness PR? I was like, it’s not just the PR. The best PR you can possibly do is to have a really, really strong product and have the product speak for itself. Right. So yeah, it’s, think it’s a very interesting time and it’s, and it’s interesting to see probably one of the first Chinese homegrown.company in AI being completely separated from the Chinese market now and operating in the West per se and now being bought up by American US company. Okay, talking about startups, I want to ask your opinion on MiniMax and Zhipu They both submitted their prospectus now. We are expecting them to go public in Hong Kong.What would you need to see before you feel comfortable owning one of these IPOs and how do you evaluate these companies as they go public?Kevin Xu (19:40) So first of all, I am actually a public market investor. So I don’t do any VC at this moment. So I’m very, very interested in how the Zhipu and Minimax listing happen. Even though as a rule, I don’t invest in IPOs because they’re quite frothy and confusing. I’m happy to wait it out. I think there are a couple of signals. And this is actually interesting as a comparison to Manus. If you look at the revenue numbers that Zhipu and Minimax have shared, they’re both in theGrace Shao (19:45) Okay.Kevin Xu (20:08) double digit USD million range, right, which is very modest. And it’s even more modest compared to their losses, which is all in the hundreds of millions of USD as far as how much money they’re losing right now as companies. And you compare that to Manus, which probably is like reasonably profitable at this point as like a hundred million dollar ARR company, not revenue ARR, but still they’re small, they’re growing and they’re probably managing their costs.because they’re not model trainers, right? Like I think Manus was very clear that we don’t build models. We don’t really have expertise in that, but we are very good at wrapping a model into a very good, trusting user experience. But Drupal and Minimax both became or started out as the model makers, which is a very expensive endeavor. So as far as what I look for as an investor is concerned,It’s very, hard. I think a path to profitability, and specifically, think EBIT profitability, so earning before interest in taxes, is going to be key for me to see how does a business ⁓ like this, which has a...I don’t know. I feel like they’re limited to the China market, which is big, but not huge, I would say. And I think MiniMax does have some consumer product similar to ChatGPT, which is going to be how they can maybe justify their higher evaluation, even though most of the revenue comes from serving up their model as a form of APIs, which is a B2B play. How do they balance those two, which are two very different go-to-market motion? It’s going to be interesting.pretty clear path or lane at this point, which is I make my models and I’m very good at serving large, older legacy enterprises and governments that is a very specific type of customer with a very specific taste, if you will. And you have to really orient your whole company to cater to that kind of customer. And Drupal kind of has cornered that market for now, at least. So that could work really well from a profitability perspective over time, even though those are very tough customers.customers to track. But the big takeaway, I think, is that the revenue is still very modest and certainly very modest compared to the large labs that we just sort of talk about willy-nilly in the US, like OpenAI, which is going to have $20 billion.in ARR by the end of this year, probably, right? Like Anthropic is projected to have five, $6 billion in ARR. These are two orders of magnitude larger than Whatchupu and Minimax has shared to the public. But the enthusiasm for investing in AI pure play is still very high in the public market. And I know Hong Kong’s IPO market has been doing very well this year and probably will continue. So that energy can be kepthopefully by these two companies because they actually need the money, right? That goes to what you were mentioning before, which is that I think if we had done this, Chad, in 2018, there will probably be multiple rounds of VC in China with USD backing that are readily available to fund the Gipu and Minimax for maybe two, three more rounds. So they don’t have to go public.Grace Shao (22:53) They need the money.Kevin Xu (23:14) They can still operate as a private company, raise more money, mag around, just like what we have been doing here in the US. But that option has basically run out of this course.right now for any Chinese VC-backed company. So they kind of have to touch the public market earlier in their life cycle for fundraising, which may not be a bad thing for organizational perspective, because you do become a more disciplined, well-run company for the most part, I think, when you become public. But it does expose you also to the public market. But they need the funding clearly, so that’s why they’re doing it.Grace Shao (23:47) Yeah, I actually just interviewed one of the leaders at Zhipu recently for the podcast and he was saying candidly, for them, it’s really about survival at this point because they’ve just run out of money. And if they don’t want to be swallowed by someone else and if there even is a desire to swallow them, because given that, you know, all the BATs we see have very, very strong labs themselves, they don’t really need to acquire a talent, new talent pool. So then there’s no way to keep going unless they go public because they need that money.But on this point on them going public and you know, actually it’s, it coincide with them trying to go global, right? A lot of them, like you said, they’re currently serving China as a market, but they are selling their model as a service to the global South, maybe for a much cheaper price than a lot of the US labs and the peers out there. How do you view that? Do you think that’s something that could potentially work out for them just by selling cheaper services compared to maybe the open AIs of the world?Kevin Xu (24:39) I think it could.Yeah, I think it could. mean, I think USAI, American AI is very expensive. Like the quality may or may not justify the premium, but it’s very expensive, right? Like we have like thousands of dollars.Grace Shao (24:44) Mm.Kevin Xu (24:51) I think the max chat GBD plan is like 200 bucks. People want like $1,000, know, no rate limit, chat GBD plans. And we’re spending a lot of money. And that’s partly goes into these revenue numbers, right? The billions that we’re talking about. Like you can think that is like an inflation almost of AI product costs here in the US for the most part. But we know that Chinese entrepreneurs are very good at reducing costs, right? They’re already released their models because the models are commodities. They’re open source. There’s not a whole lot of value capture really that happens at theGrace Shao (25:14) Hmm.Kevin Xu (25:20) autolayer. And if you can wrap that around with really good services for just throughout random examples of like a city government in Malaysia, right, or a hospital in Thailand, for example, right, these are all the kind of unsexy industries in very unsexy countries when it comes to AI adoption that we don’t ever really think about. But if they have a strategy to go after them, and I do think listing Hong Kong as opposed to on theyou know, Shanghai market, which I think could have done as well. But choosing Hong Kong is very smart because it increases their name recognition, their exposure ⁓ in that part of the world. As much as you and I talk about these companies like everybody should have heard about them eons ago, most people don’t know what these companies are. They don’t know what the differences are. They have no idea. They probably have heard of Chachi PT, but that’s about it. They probably don’t even know what Ethlopiq really is. Right. But if you can reallytap into that capital market and use the public listing as a way to raise your profile for these second tier market and second tier countries, then I think there’s a decent business to be made there.How much will it fetch a premium in the public market? I will never know. But I think that’s been a playbook for a lot of Chinese companies that were shut out from what’s called the premium markets globally, which is the US, Canada, and Western Europe. And they have to go to the so-called global south to make a living. And they’ve been able to make it work. And there’s no reason to just assume that these companies can’t make it work either.Grace Shao (26:40) Yeah.Yeah, for sure. Okay. I want to talk about sovereign AI. You’ve written a lot about sovereign AI and you’ve used South Korea as an example.Why is South Korea a champion basically in the region as for sovereign AI?Kevin Xu (27:09) I think to back up a little bit, sovereign AI is one of these things that ⁓ I’ve been really fascinated with for a better part of this year, ⁓ which is, it’s the first time I’ve seen where a major technological transformational period has beenaggressively embraced by national governments everywhere, right? Part of that has to do with Jason Huang of Nvidia just being the incredibly charismatic salesman that he is, right? Like sovereign AI, he did not come up with the term. I think it came from the EU in 2019 or something, but he really embraced it as the next wave of AI adoption. So more countries can have their own AI, which initially I thought, oh, this is just like a clever sales pitch, you know, to kind of sell more chips. But if you really think ofabout it, ⁓ all these AI models do have a way of encoding culture. Encoding not just your mainstream culture, but also your minority culture, your different languages and whatnot. And the countries have learned, I think, their lesson by being really hands-off during the first wave of the internet and especially social media, but not caring about how does technology impact their domesticsituation, if you will. You can talk about in terms in the context of Arab Springs or, you know, violence in Myanmar or just generally speaking data privacy, social media, all this sort of stuff that countries used to have just by definition a lot of control over by having sovereignty and they’re actually losing sovereignty.to the wave of technology. So with this AI coming together, this wave, more and more countries are actually exerting that notion of sovereignty without really knowing what it means, but they’re exerting it right now more aggressively than ever before. Now I picked on South Korea because sovereignty is kind of this big fuzzy word that means different things to different people. But if you use sovereignty as a proxy to talk about control, South Korea actually has probably one of the betterset of tools to exert more control over their own AI future more than other countries. Because if you really think about full stack AI from top to bottom, from land power chip models and then applications, only the US and China really have a grasp of every single layer of that stack.Grace Shao (29:24) component, yeah.⁓Kevin Xu (29:25) in diff todifferent degree, obviously, but you know that they have control over every single step, right? Every other country for the most part is a customer of one of those stacks coming from the US or coming from China, except I think for a handful of countries, South Korea being one of them because it has a very strong memory.⁓ ecosystem for chip fabrication, not for logical chip, but for memory. And high bandwidth memory is basically an exclusive South Korean national export at this point coming out of SK Hynix and Samsung. Yeah, we have some Micron over here in the US too, but the two thirds of the market is dominated by two Korean players.Grace Shao (29:46) Yeah.Mm-hmm.Kevin Xu (30:03) And then they have their pretty cool little internet ecosystem as well with Naver, with KakaoTalk. They’re all very Korea-centric. They don’t do so well outside of Korea, but inside Korea, just like how we go to China, we have to install WeChat. If you go to Korea, you have to install KakaoTalk. You have to install Naver for your map. Otherwise, you just can’t get anywhere, right? So you kind of have...Grace Shao (30:23) There is Google Map.Yeah.Kevin Xu (30:24) Exactly.So they have that set up cone coming into it. So they actually have a bunch of different good dominant controls nationally throughout all that layer. So when Jensen visited South Korea recently to sign a bunch of deals and allocated a bunch of black wall chips to different major players among these Chibos, I just thought this is like an actual manifestation of a South Korean sovereign AI at play. Now they’re still using American chips, but part of that American chip is fused with South Korea made memory.And that gives them a lot more say, at least, to sovereignty of the AI application that they’re hoping to adopt. And South Korea is just very digitally forward, I think, in general. It’s one of the most digitally connected society, period, of any country in the world. And so ⁓ I think they have a good shot at actually making sovereignty real in the AI era.Grace Shao (31:03) Yeah.It’s interesting because I just went to Korea I think earlier in the year and I was talking to investors on the ground and they were saying that South Korean startups are actually a lot more, again going back to our point, non-geopolitically driven or minded and a lot more agnostic about which kind of, what countries models they use. However, for the country itself right now, the government, they’re still pushing US models forward. And I think it’s really interesting to see to your point likeThey actually have such a small but closed ecosystem in the digital infrastructure. Like everything is with, they don’t use American apps like for social media. They don’t use Chinese apps for social media. They’re actually completely independent. So it would be an interesting kind of case studies to follow through with, I think. When we look at sovereign AI and I look at, know, projects like Stargate, is that something like a de facto US sovereign AI project? Like, how do we understand that?Kevin Xu (32:03) That’s how I understand it. I think Stargate is, first of all, for people who don’t follow this stuff as closely, is this brainchild from OpenAI to build these massive multi-gigawatt data center, not just in the United States anymore, but actually throughout the world, to support its global multi-trillion dollar ambition. And it’s in countries that are willing to be on Team America. So in a way, it’s a sovereign extensionGrace Shao (32:05) Mm-hmm.Kevin Xu (32:31) of American AI in a way is also a reduction of sovereignty in whichever country is willing to receive American AI and be a proxy of American AI, right? And we have a few different sites already announced. We have ones in Argentina, in the UAE, in Norway. I think these are the ones outside the US Stargate projects, maybe perhaps India as well. And that is the most aggressive.expression of American sovereign AI and the most explicit one as well. And the US government is very honest about this as well. Like they want to promote and literally sell the American stack to countries around the world that want to buy American projects. It’s basically like a bigknow, White House driven go to market strategy, right? Where the content of the product is actually open AI, Anthropic, Nvidia chips, Oracle, construction, and all the American kind of major companies that come together into literally a package, right? And then we want to sell that abroad to different countries around the world, including the global South as well. And I think that’s one of the things that a little bit of a shift geopolitically is that the US is no longergiving up the global south as this also ran that it is no longer paying attention to in the way that China has been paying very close attention to for two decades at this point. It’s no longer willing to surrender that part of the world commercially. And Stargate and AI export program is actually a way to express that re-interest in those regions, if we will. And Stargate is just kind of the tip of the iceberg there.Grace Shao (33:56) Mm-hmm.I think to talk about sovereign AI, we have to talk about Middle East and it’s something I really know nothing about. I was really fascinated by your recent article and your series in sovereign AI. So first for listeners, can you tell us about your trip? You just got back from Abu Dhabi, I think a week or two ago, you wrote a really insightful long piece on just how the Middle East is building out their AI strategy. And you talked about it as a region, also kind of breaking it down the whole, looking at the Gulf separately, the UAE, the Saudi Arabia, Qatar, Bahrain, each of the...AI strategies, right? Can you kind of walk us through, first of all, why did you go? What was the event for? And then just some of the high level takeaways from that trip.Kevin Xu (34:44) So I went to that trip from the exact same position that you are now, which is that I’ve never been to the region. I’ve heard a lot of things about the region. Just in the AI conversation alone, we’ve had major announcements and deals being signed by Saudi Arabia, by the UAE, with the United States. We know Chinese tech have been in that region for a very long time as well. There a lot of robot taxi Chinese companies that are deploying their self-driving vehicles on the ground as we speak. So it’s a region that I’ve been really wantingto go if I get the chance to go for a long time. And just by happenstance, I was invited to be part of a delegation among other Washington, D.C. think tankers to go to visit the UAE for a week. So we are an American delegation, right? So that’s important context for you to know. If you were to read the post that I wrote and understand what I was trying to convey and how I learned things, we spent a whole week in both Abu Dhabi and in Dubai meeting with pretty much everybodythat has a hand in its AI future, from government officials to investors to all the funds that you have heard of. And the major takeaway for me was, first of all, just to see stuff on the ground, which is thatThey are very, ⁓ from an AI perspective in particular, just take away the other stuff for now, from an AI perspective, they’re very much in Team America’s camp. They really want to be building UAE Stargate. That’s one of the very few Stargate projects outside the United States that has actually broken ground. Like there are actual buildings that have been built in the desert.⁓ ready to receive NVIDIA BlackWall GPUs if expert control were to be permitted from the US side to let them buy as many as they would like to buy. So that’s number one. I think number two, they are in this very interesting geopolitical position as a very tiny country of 10 million people where they don’t want to actually be Switzerland. They’re not neutral. That was the message that I received from a lot of people that they have a point of view on where they want to be in this globala big game of AI, of geopolitical influence, which is that they can vote for one country or one side, but they also have their opinion to build a society of their own. That’s a very modern Arabic.you know, society, I think there’s a lot of stereotypes to, know, how does it, what is it like to be in a be a woman in the Middle East? What is it like to operate in the Middle East? Lots of cultural stereotypes that they want to debunk. It’s Dubai is one of the most modern cities I’ve ever ever been to. Right. And that is kind of the cultural takeaway that they want us to have. And then lastly, when it comes to this US China conversation, frankly, they’re a little bit tired that they always get mentioned in that context. Right. Every time the US official goes to the UAE is aboutWhat are you doing with China? then, know, presumably when the Chinese official visits, they’re also like, what are you all doing with the Americans? But they want to be seen on their own term. And they certainly have the wealth to do so.as well. So it’s a fascinating region and I think they’re playing both sides very well. I call them the swing vote of the global AI competition. They can swing one way, can swing the other, but they have a lot of leverage in this conversation because they need to have the best of both worlds to feel an economy that is in the desert that literally grows nothing.So they have to export import, sorry, they have to import basically everything from talent, from food, from vegetables, from, you know, the only thing they have is energy coming out of the ground, oil, but they’re trying to diversify away from that, which is the only point where they’re investing all this technology stuff in the first place. And that has been happening for 20 years at this point. So it’s not like a Chad GBT moment thing per se. So a lot of takeaway there, but happy to answer more questions because it’s a trip that I’m still processing, to be honest, because that was first time in the region, had a lot ofcoming at me and I’m trying to still come to terms with ⁓ what I understand now but also what I still don’t understand even though I just went there.Grace Shao (38:40) Yeah, I think that that’s super interesting. And I’ve been really fascinated by the region as well. Actually, we were just talking about this offline. A lot of people in Hong Kong are now being recruited over on the point of talent. And I think, you know, as a lot of these countries have huge sovereign funds, they’re looking for top tier investor talent to go to whether it’s UAE or Saudi or Qatar to really deploy that capital, whether it’s an AI or not. And it’s really interesting kind of to see howYou mentioned they have, what, UAE has 10 million people, but 90 % of that is actually foreign workers, including laborers, as well as knowledge workers. And people kind of forget that actually these are extremely wealthy countries per GB per capita. So I think it’s interesting to hear that they don’t really want to be put in the middle as a China camp or a US camp country now. Similarly to Singapore, where we also talked about, know, like Singapore isTiny small, you know peninsula has actually really made it work from themselves and Pretty much have to import everything from groceries and labor is sometimes from Malaysia and even energy to talent from around the world and now mostly China It’s kind of like in that sense not a Switzerland like Singapore right like you said in your article. I want to understand better actually How do we understand the sovereign funds behind?these investment funds are investing in AI because it actually is so different from private capital in the US and even how Chinese capital is structured.Kevin Xu (40:06) The way I-the sovereign fund in the UAE in particular, know, just that part of the Middle East have not been to Saudi Arabia or anything. Obviously Saudi Arabia is a major, major player as well. So is Qatar, which actually announced their own AI initiative while we were in the UAE as part of the Doha Forum. So there’s a lot of, let’s also call it co-op petition as well, among the Middle Eastern countries as well. Like they’re presented as this sort of monolith sometimes, but there’s actually a lotGrace Shao (40:23) Exactly,Kevin Xu (40:37) of rivalry or friendly competition between, in particular, these three Middle Eastern golf countries, Saudi Arabia, UAE, and Qatar. Now, the UAE sovereign wealth fund in particular, again, we met with everybody there. so their strategic purpose is, of course, to diversify away from oil wealth.Right, but that’s easier said than done. What do you do when you have all this money? From selling oil that you know, it’s gonna run out at some point or you don’t want to be overly dependent on this one source of wealth, right? So Mubarak is their kind of marquee sovereign wealth fund that plays very actively in the world of technology investing and they’ve been Investing for 20 plus years around the world. They’ve had offices in China in South Korea in Brazilobviously in the United States, in Europe for many, many, years. They’ve been placing bets and serving mostly as LPs to local VC firms fora long time. They’ve also bought a global foundry, which is the chip manufacturing plant, similar to TSMC. But you know, that was kind of a spin out out of AMD, I believe, back in the days. So they’ve kind of placed their bet in the chip ecosystem, again, long before AI was a thing. Now, that doesn’t mean they’re all successful, because the diversification justification is very different from ABC, who is motivated to generate the largestfinancial outcome right per fund per fund and they actually did understand more recently why that’s not such a good model which is directly related to your point about Hong Kong professionals finance professionals being recruitedto go to the UAE because they started this new fund called MGX, which is basically more of classic VC fund that has all the incentive structures of a Silicon Valley VC firm like a Sequoia or a 16Z. Mubadala is one of the anchor GPs, but they’re raising money from around the world just like a normal VC would because they need to attract the best talent, which they actually could not if you just run asovereign wealth fund because sovereign wealth fund is kind of like a quasi government institution, right? They’re still kind of government employees at the end of the day. They don’t get a huge carry or a payout because one of their funds hit it out of the park and got less than the NASDAQ. They’re just kind of collecting their paycheck, right? They’re more like a pension fund manager. And that doesn’t get you the best, most hungry, I don’t know, money.making talent from London or Hong Kong or wherever. So they’re just very recently started to restructure that because it’s an evolution of sovereign wealth fund being managed, one, to diversify and then to get into the best technology and then to actually generate a good return and get the best talent, which is really a long-term play because if they can lock down the best talent from Hong Kong for a decade or two to live in Dubai, to live in Abu Dhabi, then that is a long game that they can, again, supplant this 10 million people that needs to be constantly replenished.with better talent and more diversified talent. So the sovereign wealth fund, the game is very complex, I think, and they probably played it better than most people that I’ve seen coming out of any sovereign wealth fund. Singapore sovereign wealth fund is very sophisticated as well, but that took a long time to evolve GIC and Tomasic.Grace Shao (43:49) Yeah.Yes, yes. That’s really interesting context. I think I have one last question for you on Middle East, just given the time constraint, but I would love to talk more about this offline. If you had one non-consensus bet on the Middle East and how it may shape AI globally in the next few years, what would it be? Like, how do we understand the Middle East role going forward, especially amongst this US-China co-petition?Kevin Xu (44:18) I think I was skeptical going into the trip that it’s going to be a region that actually would matter because there so many data centers being built everywhere. But coming out of it, ⁓ I think there is a good chance that the UAE Stargate will house a significant amount of compute for not just that region, but for the entire world.First, because its energy is abundant. Second, its construction force, which is something that we did not talk about explicitly. They have 1.5 million construction workers. So 15 % of the population in the UAE is constructing something. They wake up, they’re building something. It could be a hotel, it could be a resort, or it could be a data center. That is something that we’re actually very...Grace Shao (44:59) these are mostly workers from abroad, right? From India, Pakistan, Philippines. Yeah.Kevin Xu (45:02) These are almost, these are entirely workers from abroad, right? These are Pakistanis,a lot of South Asians who are there on workers visa. So they’re not, you know, living some glamorous life. They’re just a construction worker life, right? And there are a lot of kind of like issues with that approach, if you think about it. But as far as the capacity is concerned, they’re able to really build stuff faster than just about any country that I’ve seen. And as the United States hits its challenges, I think, when it comes to labor,when it comes to energy capacity and I think will also become a domestic political issue very very soon especially this upcoming year with the midterm election that could grind a lot of the pace to a bit of a halt and the UAE is ready to risk kind ofreceive all that chips that are being made in Taiwan. And I think that will really be something that people haven’t really thought about as far as where their computer will actually physically live, which really will bring again the sovereign AI story of the UAE toto life because it’s not just another talking point anymore. They actually have a significant amount of compute that could be used for training models and it can also service a bunch of the region over there because the telecommunication cable between the UAE and say India, for example, or Pakistani, the speed there is like 30 or sub 30 milliseconds, which is super, super fast. So you can actually serve a bunch of users from the UAE to India if you’re okay with that kind of, you know, data center set up.So that’s something that I think people are probably still sleeping on. We may see that becoming a more real just in another 12 months or so.Grace Shao (46:39) Interesting. Kevin, you’re so knowledgeable and everything. I love reading your work and I just really enjoy this conversation. I have one last question for you, which is a question I ask every single guest that comes on the podcast. What is one differentiated view you hold? Non-consensus, something maybe even controversial that you truly believe in that maybe your peers don’t?Kevin Xu (46:57) I think, I’ll share two, but they’re interconnected. Obviously they’re really one, but in two parts. One is that I think there’s a consensus that China AI, AI in China is diffusing better than the US. I think from an economic perspective, from an economic impact perspective, that is actually not true. If you just compare the revenue number,between Gipu and Minimax to any lab that we have here in the US. It’s peanuts, right? Now you can say we have a bit of a token price inflation over here in the US, as I’ve admittedly mentioned during our conversation, but it’s not 100x premium as far as like that delta is concerned. So there’s actually a lot of economic ⁓ value being captured here in the US just by the diffusion pace that we’ve been able to push out here alone.So that’s sort of a non-consensus thing, the one. And the other thing that is related is that because the pace of diffusion in China is a bit more up and down the stack, you know, not just in knowledge worker, but in factories, in on the road with self-driving and in robotics and whatnot, let’s just assume all these will just kind of continue at pace faster than any other country in the world. Then China will also be the one countryThat has to deal with all the social ramifications of AI before any other country in the world So this is a very interesting moment where the Chinese government and regulators will have to lead the world Into this kind of dark space as we’re all filling out what the hell this AI is gonna do Before anybody else and I’m really interested to wait to see how much they’re willing to share their learningtheir failures, their successes from a rulemaking perspective? And also, how humble will the European regulators and the American regulators be willing to learn from the Chinese failures so we don’t screw up too much in our own backyard?That will be something that I think will happen for sure, but it could really determine the direction of where all this is going. And we kind of saw a little bit of that with the most recent regulation coming out of China when it comes to regulating the chatbots. It’s much more prescriptive than the usual list of harms when it comes to data privacy and whatnot. It touched very specific use cases, like if a chatbot is going to talk a lot about, you know,Grace Shao (49:02) What’s you say?Kevin Xu (49:16) the giving mental health advice or all these much more personal use cases that could lead to self-harm the regulators in China is having a very particular point of view on how this should be Diffused in its society whether that lesson good or bad gets learned here in the US and elsewhere in the world is Gonna be interesting but China will have to lead on this front ⁓ Which is a position that I don’t think the Chinese regulators are used toGrace Shao (49:42) Even expected,Kevin Xu (49:42) ⁓ up to this point.Yeah, they’re used to learning from outside. They’re very good at absorbing the best rules from Europe and the US to bulk up their own regulatory capacity and knowledge. But this could be the one thing where they will be the first to step into the abyss and they have to help us get out of it.Grace Shao (49:59) I think a really, really interesting point. And I actually been thinking about this as well. To your point, I diffusion in China is so much more obvious to the human naked eye because it’s seen through consumer usage, through just the rampant digital infrastructure buildup that we’ve seen in China. So everyone, like you said, from random auntie to knowledge workers will be using AI. But the actual capital gain, the real money has not been proven to be greater than...than the US and already we can see that from just IPOs like MiniMax and Zhipu And I think the regulation that you were talking about actually came out interestingly right after MiniMax and Jhipu actually released their prospectus to the public. So it’s like, I think regulators are really keeping a keen eye and a hand on it and trying to see what could potentially happen. We speak to people in China practicing AI, like the actual builders and the scientists, they say, there’s less of a discussion about this like.Doomerism kind of view people are taking more pragmatic view, you know, people are really focused on technological advancements less about societal implications Yes, I kind of believe that being probably the case given that you know in China last 20 years people really saw technology as a Path to economic prosperity, but however, I think what your point is is really interesting is that actually this time they can’t see what happens how the US regulates by techthey have to do and start themselves, right? So that’s a really interesting point. I actually will think about that a bit more as well. ⁓ Thank you, Kevin.Kevin Xu (51:29) Yeah, there’s a 100% chance that China will have to be the first country to lay off a bunch of delivery drivers and, know, ride sharing drivers if robot taxi becomes a thing, right? What will Wuhan do? I think everybody else in the world will want to know when that happens. Yeah.Grace Shao (51:46) Yeah, yeah, especially whenembodied AI becomes more of a but okay on this point I wonder your thoughts on this because When we go to China, it’s really interesting you have these random jobs that are like placed for sure not for like actual practical reason like you know those aunties who sit in elevators and press the button for you or Like a uncle who sits there like an older kind of larger man who sits outside the parking lot who just pressed the toll button for you likeKevin Xu (52:03) Mm-hmm.That’s Right. That’s right.Grace Shao (52:13) Thesejobs frankly are not needed, but they’re implemented I think for societal harmony purposes because you need employment. You need to give these frankly not very skilled laborers a job. So if you’re gonna push for embodied AI in China and these physical, whether it’s robots or whatnot, are gonna replace a lot of these lower skilled jobs, what’s gonna happen to society? Do you think they’ll actually?implemented at mass or do you think they would actually take a more cautious decision?Kevin Xu (52:43) My read on that is they will be very, very cautious, which again goes to the non-consensus view that I just shared on the diffusion narrative about China and AI. Right now, the consensus is that, oh, China’s diffusion is so much faster. They’re going to push all this AI. We’re screwed here in the US. But really, there is a very good human reason to not do that.⁓ You know, this is not exactly public knowledge, so I won’t cite it. But if you look at the pace of deployment of the self-driving companies operating in China alone, right? know, you Baidu, you have Pony, you have Rewrite, you have some of smaller players. It is, they’re all born there. They have very good regulatory environment to experiment and develop their technology. But they’re actually throttled by local permit capacity.on a city by city basis as far as how many of these cars can they actually deploy on the road? Because it’s not a free for all at all. Every city is looking at the numbers and be like, okay, if we actually do this tomorrow, like let the floodgate open because the technology is actually really, really good already. And we already know the Chinese, yeah, well the Chinese OEMs can pump them out really quickly, right? I think that the safety concerns actually getting really, really good. But what would the delivery drivers do?Grace Shao (53:49) It’s not a safety concern.Kevin Xu (53:59) What would the DD drivers do? So there is this toggling already between how much are the government willing to let this technology loose versus taking care of the aunties who pressing the buttons and the dachu who’s letting you into the parking lot because there has to be a pathway there. It’s just not obviously a subsidy program, but it’s clearly a government funded economically irrational employment program.Right? Like the only corollary we have in the US is the greeters at Walmart stores. I’ve never been to a Walmart super center. There’s like this person who just says hi to you and you walk in and you get your Walmart stuff. Like does that person need to exist? It’s of Walmart’s premium user experience for shopping there. But we don’t have as much of that here in the US, but we certainly have a little bit of that too. Right? So again, China is going to hit that at scale.Grace Shao (54:23) Yeah.It’s part of user experience, Kevin. They want you to feel welcome.Kevin Xu (54:45) before any other country. And they’re trying to figure out the right balance right now as we speak, but we don’t really have a good sense, at least from the outside, of what are the rationales, can we learn from that, can they share more of the thinking, so we can all kind of benefit from that, from a rulemaking perspective.Grace Shao (55:01) Yeah. And you saw that with the urbanization demand, like what, 10 years ago, we saw a huge rise of young men from rural areas moved to urban cities to become delivery workers, whether food delivery or package delivery courier, that created a lot of economic gain for the country. And then when COVID hit, it was crazy. A lot of people got laid off from their white collar jobs. And then you saw a huge increase of essentially Chinese Uber ride drivers.Kevin Xu (55:27) That’s right.Grace Shao (55:29) So all of a sudden people all became drivers and there’s a huge oversupply of riders and now you can call a DD and any car would shut up within like a minute. It will be interesting where would these people go if you’re gonna introduce all these self-driving cars, self-driving delivery man, whatnot. It will be interesting because that makes up a huge part of the urban economy right now. Yeah.Kevin Xu (55:43) Yeah.That’s right. That’s right.And you know, one approach is just that you don’t, right? You just say, okay, we know we have the tech, you can export to the UAE all you want, which though they’re doing really well in the UAE, the Chinese are all with taxi companies. But at home, you’re going to pace yourself because we have a lot of people who are going to get really, really upset if this thing gets unleashed tomorrow, which it can. And that kind of goes against the whole China that just defuses everything because China loves AI sort of narrative.Grace Shao (56:16) Yeah, interesting. Thank you again, Kevin. Really appreciate your time and your insights.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  EVs taking on AI OS and the Delivery War. The Chinese Tech Winners Beyond BAT with Alan Zhang 06.01.2026 49λ

  In this episode, I sit down with Alan Zhang (Principal & Portfolio Manager at Ox Capital Management) to map China’s tech landscape through an investor’s lens. We break down how Alibaba, Tencent, and ByteDance are approaching AI, and why the “AI OS” is the real endgame. Finally, we analyze what’s changing in China’s consumer internet, EV ecosystem, and embodied AI pipeline. We also unpack China’s delivery wars (Alibaba vs Meituan vs JD), why quick commerce is structurally different from traditional e-commerce, and how markets price geopolitical risk into China tech valuations.Alan Zhang is a Principal and Portfolio Manager at Ox Capital Management, a boutique investment firm focused on emerging market equities that he co-founded in 2021. At OxCap, Alan leads investments across Asia; previously, he spent years as an investment analyst on the Asia team at Platinum Asset Management.He studied Actuarial Science and Commerce at the University of New South Wales, and he’s even taught advanced econometrics. So if you like the intersection of fundamentals, market structure, and Asia platform businesses, well then, this one’s for you.In today’s world, there’s no shortage of information. Knowledge is abundant, perspectives are everywhere. But true insight doesn’t come from access alone—it comes from differentiated understanding. It’s the ability to piece together scattered signals, cut through the noise and clutter, and form a clear, original perspective on a situation, a trend, a business, or a person. That’s what makes understanding powerful.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.For more information on the podcast series, see here.Chapters 01:34 Alan’s background: quant → Asia equities03:11 US vs China AI: frontier vs “two-legged” approach05:25 “Uninvestable” China and what changed07:31 Beyond BAT: Xiaomi, Meituan, Mindray, MicroPort09:24 BAT AI strategies and the AI OS thesis13:45 Tencent: tools, data, distribution, and model strategy16:33 AI-native phones: ByteDance × ZTE and what’s next26:51 China EV landscape: BYD, Huawei, Xiaomi, Zeekr31:28 Why phone OEMs can compete in EVs34:16 Embodied AI: robotics parts, redundancy, and Unitree39:38 Valuation + geopolitics: why Asia tech trades discounted41:53 China delivery wars: subsidies, quick commerce, Meituan’s edge50:27 12–18 month predictions + what investors miss (healthcare)AI-Generated TranscriptGrace Shao (00:00)In today’s world, there’s no shortage of information. Knowledge is abundant. Perspectives are everywhere. But true insight doesn’t come from access alone. It comes from differentiated understanding — the ability to piece together scattered signals, cut through the noise and clutter, and form a clear, original perspective on a situation, a trend, a business, or a person. That’s what makes understanding powerful.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend — someone who can help us see things differently.So today, joining me is Alan Zhang. And I’m Grace Shao. Alan, really excited to have you. I’m excited about today’s conversation because we’re going to get into the investor’s perspective on Asia tech and emerging markets — with a proper markets-and-math backbone.Alan Zhang is Principal and Portfolio Manager at Ox Capital Management, a boutique investment firm focused on emerging market equities that he co-founded in 2021. At OxCap, Alan leads investments across Asia. Before that, he spent years as an investment analyst on the Asia team at Platinum Asset Management. He studied actuarial science and commerce at the University of New South Wales, and he’s even taught advanced econometrics.So if you like the intersection of fundamentals, market structure, and Asia platform businesses, this episode is for you. Alan, welcome.Alan Zhang (01:31)Thank you, Grace. Pleasure to be here.Grace Shao (01:34)Alan, to start, why don’t you tell us about yourself — your background — and what it is that you cover now?Alan Zhang (01:40)I grew up partly in Hong Kong, mainland China — Shenzhen particularly — and in Australia. I spent close to a decade in Australia doing my schooling and education, and worked for a firm called Platinum Asset Management, then co-founded Ox Capital with Joseph Lai.I studied actuarial science, so I’ve had a lot of experience manipulating numbers, cleaning up data — and that helped me tremendously in public equities. Nowadays there’s no shortage of financial data, and the ability to understand them — and the intent behind them — is crucial to investing.Grace Shao (02:34)Yeah, yeah.Alan Zhang (02:46)At Ox Capital, we also built a tool called the Mode Model, which distills more than a million financial data points from various sources to help us understand our coverage region a lot more. In terms of my coverage, I build quant models, I look at equities, and I also help with portfolio positioning based on macroeconomics in Asia.Grace Shao (03:11)That’s interesting because you started off in quant, but now you’re looking at equities — the fundamentals, right? You’re covering a lot of ADRs, and a lot of China’s big tech.Let’s talk about that. What is the China big tech internet ecosystem looking like right now? How does it compare to the US?Alan Zhang (03:20)In the US, they are focusing more on frontier models, while Chinese companies are taking more of a two-legged approach — tackling AI with different approaches. The US has invested a lot of resources into advancing frontier models. On one hand, we see successful cases like Gemini, Anthropic, and OpenAI, while we also see a lot of AI subscriptions cutting their prices by more than 90% in the last few years.If you remember in 2023 and 2024, many subscriptions were priced at a few hundred — some over $1,000 a month — based on investment assumptions. Now they’re cutting prices to sub-$100 a month. Some may never make their money back based on those assumptions, but it’s not being discussed today because the benefit of AI far outweighs that blip, and large-cap companies are investing enough to offset the impact.If we look at China, they haven’t gone through this episode — and I don’t think they will. Anyone who looks at Asia understands Asian users will never assume people will pay over $1,000 a month for subscriptions. China is working on frontier models, applications, and infrastructure at the same time.In summary, China is still the runner-up, but they’re developing AI in a more balanced manner. And it’s also good to see the US pivoting — in the recent 12 months, we’re seeing more US companies investing in software and applications rather than just frontier models.Grace Shao (05:25)China was deemed uninvestable, especially for Western investors. Your fund is based in Australia and Hong Kong, and your LPs are non-Chinese. For public investors who want exposure to China’s AI upside — what are they looking at? What are they thinking?Alan Zhang (05:46)Usually the big tech. China went through the property adjustment and the antitrust campaign in the internet space. It was painful — people called it uninvestable because they couldn’t see new growth drivers. And if they could, they were too insignificant compared to the two most important industries at the time: internet tech and property, which were both recalibrating.But things are different now because investors can see new growth drivers scaling up. In hindsight, these adjustments also helped innovation: talent that dreamed of landing a job at Meituan, Tencent, Alibaba went to smaller firms or startups; capital that made easy money in real estate went to new areas.Economic transformation is still a work in progress, and investing in China becomes more attractive if we see AI, consumption, and advanced manufacturing play a bigger role. We’re still in that phase. But we’re glad to see some companies bottoming out and making progress under the current setup.Grace Shao (07:19)In a pragmatic way, does that mean we’re looking at BAT? What companies should we be looking at for exposure to Chinese AI and economic transformation?Alan Zhang (07:31)Besides Alibaba and Tencent, people should look at relatively smaller cap — but still large-cap — companies like Xiaomi and Meituan. And also industries outside the internet. For example, Mindray in healthcare, or MicroPort in surgical robotics — they can implement AI into their products and make their portfolio more attractive.Grace Shao (07:41)When we chatted offline, you said a lot of companies are overlooked. Beyond BAT — what are some “1.5 tier” or “second-tier” companies that are huge by market cap but not well known in the West?Alan Zhang (08:09)People will naturally see them more over time. Tencent and Alibaba were making active efforts overseas; now as the market matures, more companies are going global. If I’m on a roadshow, people ask about Keeta, which is a subsidiary of Meituan. Xiaomi is opening more stores in Europe — even Africa and South America. People will naturally see them more.If you come to China and compare what’s here to where you live, you’ll see a clear difference.Grace Shao (09:24)Let’s double click on BAT — Alibaba, Tencent, and ByteDance. At a high level, how do you compare their AI strategies? Are they playing the same game, or different playbooks?Alan Zhang (09:52)Same, but different. They’re all investing heavily in frontier models and infrastructure. Ultimately, they all want to build the AI OS people will use. The DoorDash–OpenAI collaboration was a good example of what AI and a commerce company can do. Whether it’s an app within an app or an app within a phone — that’s still an open question.Alibaba is e-commerce and cloud. They have to build a competitive model or their cloud becomes commoditized. Tencent is a platform — they build tools. In LLMs or AGI, late movers can have an advantage because users may be indifferent as long as security and usability are similar. ByteDance, as a private company with strong feed algorithms, has been AI-native for a long time — even back in 2018 they were investing heavily in AI and user intent.So they’re all trying to build an AI OS for users, just from different starting points.Grace Shao (12:29)I love that framing — I’ve been writing that 2026 is about the AI OS. Tencent has signaled they’ll double down on LLMs. It’ll be interesting to see whether late-mover advantage shows up — and whether they need to spend less on pure infra.How should we think about Tencent’s positioning? They’re late on LLMs, but AI is already integrated across touch points — WeChat, gaming, fintech, mini programs. Should they continue using open-source models like DeepSeek, or focus on proprietary models like Alibaba integrating Qwen?Alan Zhang (13:45)They’ll do both. With Yao Shunyu reporting to Martin Lau, they’ll try to build their own model like every tech giant. At the same time, Tencent’s bread and butter is building tools — AI tools to help merchants and users and improve the experience.Whether it’s an app within an app or an app on a physical phone — like the Doubao phone we saw — Tencent has the ingredients: ecosystem, quality data, and distribution.Grace Shao (14:37)When you say “building tools,” how is that different from Alibaba building tools for businesses? And how is that different from ByteDance’s “app factory” approach?Alan Zhang (15:10)One example: in WeChat’s input bar, if you long press, you can translate. People type in their own language and WeChat translates to the recipient.I also visited their AI showroom recently. They showed mapping genetic pools and building a genetic bank for seeds and animals — they have quality data. They can also build full simulators for flights and cockpits — one of only a few companies that can do that. They’re investing in spatial intelligence and data banks, and building tools inside WeChat.I think it’s only a matter of time before they move more properly into e-commerce and release something like what DoorDash and OpenAI shipped.Grace Shao (16:33)On hardware — can we talk about ByteDance and ZTE’s partnership? ByteDance worked with ZTE and launched an AI-native operating system on a ZTE phone. Instead of building their own phone, they partnered with OEMs. What do you make of that?Alan Zhang (17:11)As a user, I looked forward to it. A product like this may take longer to be widely available because it disrupts a lot of vested interests. But the trend is inevitable — AI OS will be valuable in ways we can’t even measure.This is what I envision for Xiaomi and Tencent too. Companies like these — and Apple — are planning for that day, but they’ll move when stakeholders are ready. OEMs have the protocols to make it happen. Tencent also has content and intent — ads revenue — plus distribution. Tencent and Xiaomi will try to tackle this new market.Grace Shao (18:13)Is ByteDance moving faster because it’s private? Xiaomi and Tencent are public companies — does that slow them down?Alan Zhang (18:29)Absolutely. ByteDance can try something new; if it fails, it doesn’t impact the core. If Tencent or Xiaomi do this, they can agitate business partners and users.Grace Shao (19:10)For an American audience, is there an apples-to-apples comparison to US peers?Alan Zhang (19:32)It’s difficult. These companies are mature and make decisions based on their own opportunity sets. In many spaces, Chinese companies are leading, while the US is still exploring new frontiers. Tencent has been relatively quiet until recently, and they work quietly with industries to understand how their AI stack helps.In 2015, Tencent founded a learning program called Tencent X — “X” stands for another Tencent. They work with business schools, bring entrepreneurs and business leaders to site visits and exchanges, and use the process to understand how to develop their stack to empower Chinese industries. A successful example was Pinduoduo — through this program, they found Colin Huang and supported the company through traffic. Tencent can find more companies like this in their own way.Grace Shao (20:46)[Connection drop]Grace Shao (21:04)Could you restart that sentence?Alan Zhang (21:08)[Repeats Tencent X explanation]Grace Shao (22:07)Looking at 2026 — what consumer AI applications might look different? Any sprouts inside super apps that people aren’t noticing yet?Alan Zhang (23:07)2026 will likely be an interpolation of 2025. I don’t expect a completely new form factor. Most Chinese companies are already super apps, boundaries are ambiguous, and they’re fighting for the same consumer pockets.But ads revenue will shift. Previously, ecosystems charged a lot for ads because of captive customers. With AI, people are reconsidering how they use apps — budgets will relocate to new apps.Grace Shao (24:20)On infrastructure: it feels like everyone is shipping models — not just BAT and the “four tigers,” but also Kuaishou, Meituan, Xiaomi, even EV players. Why?Alan Zhang (24:58)They have enough users, and AI improves experience and broadens reach. For example, older users didn’t use search much, but with AI they can adopt faster. AI makes products more interactive and easier to use.EV companies want more engaging products. Cars are becoming commoditized, so they invest in infotainment and ecosystems. That’s why every sizable Chinese company will try to build a model. And we’re still in the investment phase — nobody knows who wins, so everyone tries. It’s not as expensive as it sounds.Grace Shao (26:25)Isn’t it costly for EV companies?Alan Zhang (26:32)It’s costly, but a lot of money is spent on chips research and manufacturing. The LLM itself isn’t as expensive as people imagine.Grace Shao (26:51)Let’s double click on EVs. Who are the biggest players in China beyond BYD and Zeekr?Alan Zhang (26:55)BYD and Huawei. Emerging ones: Xiaomi and Zeekr.Grace Shao (27:15)How do you position them?Alan Zhang (27:21)Xiaomi’s selling point is ecosystem. You can call “Xiao Ai Tong Xue” — the voice assistant — to operate devices through the ecosystem, especially with HyperOS 3.BYD’s advantage is manufacturing — they can build similar-quality cars cheaper through supply chain management.Huawei has HarmonyOS and strong brand equity — customers pay up, so they can stack a more luxurious experience into the car.Grace Shao (28:15)How does that compare to “luxury EVs” like Nio — are they still relevant?Alan Zhang (28:24)They’re still relevant. Li Auto is more family-oriented than luxury. Nio targets younger consumers who want the driving experience. Huawei’s models skew more toward corporate executives and founders — generally 40 and above.Grace Shao (29:08)So there’s a shift — five years ago it was BYD, Nio, Xpeng, Li Auto; now Xiaomi and Huawei are making strides because of AI operating systems. Is that right?Alan Zhang (29:28)Yes. China’s auto market has many brands and licenses, no shortage of production capacity — and there’s overcapacity. The “anti-involution” campaign has targeted autos. The industry is commoditized, so companies need differentiated advantage. Xiaomi and Huawei have ecosystems; BYD differentiates through cost and can scale domestically and overseas.Grace Shao (30:41)Why are Xiaomi and Huawei able to lead? Does that mean EV-first companies become less competitive?Alan Zhang (31:28)EVs have fewer parts than ICE cars. Historically you needed over 10,000 parts; now EVs might have a few hundred to just over a thousand. You can break it into powertrain, battery, chassis, and battery management — and the rest is non-core. Many parts are commoditized except the battery and system.Xiaomi and Huawei can repurpose capabilities from phones: chips, screens, packaging. Xiaomi can repackage Qualcomm chips and repurpose them to be auto-grade; Huawei can do similar. Cars also have bigger screens than phones — manufacturing capability transfers.EV-first companies like Nio, Xpeng, and Li Auto spend on manufacturing and also on chips, because their bigger vision is robotics. They’ve said chips for EVs alone wouldn’t pay back — the bigger scheme is robotics.Grace Shao (34:16)So in embodied AI: you have Unitree, “Galabots,” UBTECH; you have EVs; you have Xiaomi/Huawei tech stacks. Who wins? Is it just cost and price?Alan Zhang (34:54)Cost, price, and redundancy for physical movement. Even traditional automation companies like Inovance are building robots. A robot shares parts with EVs — optics, gears, batteries — but also has new parts like PLC controllers where you need redundancy. On these fronts, many are on a level playing field.Grace Shao (36:12)Do Chinese EV firms have an edge in spatial intelligence, or is it mainly cost?Alan Zhang (36:21)China is still runner-up in spatial intelligence and will spend time to catch up. But China has a short feedback loop: optical components and supply chain are local; ideas can turn into products quickly and iterate fast. Not an advantage yet, but not far behind.On who wins: too early to say. Unitree is the one that can make a more agile robot and do more stunts than other players.Grace Shao (37:42)Where does AI show up in embodied systems — is it just visible “smart” functions, or more invisible?Alan Zhang (38:19)Besides user experience, AI processes many parameters in the background. With enough computing, embodied AI can make simultaneous decisions — what to move and what not to move. Humans blink, walk, and raise hands at once; without AI it’s harder for robots to act like that. With AI, robots can handle more parameters and make simultaneous moves.Grace Shao (39:38)How do you price geopolitical risk into valuation positioning? Export controls, trade wars, domestic regulation — how should investors look at China?Alan Zhang (40:17)The market is already pricing a discount. Asia tech trades at a discount to US peers — Samsung and SK Hynix versus Micron; BAT versus the Magnificent Seven. Tools may be less available, which can slow advancement, but it’s also encouraging to see alternate solutions like DeepSeek. Over time companies can become more technologically independent.For large caps, investors may feel safer sizing up. For small caps, we start small and see how it plays out. Entrepreneurs are agile and prepare for change.Grace Shao (41:53)A reader question: China’s delivery wars. Alibaba vs Meituan — subsidies, vouchers — why is this happening now?Alan Zhang (43:43)Meituan has led quick commerce — 30-minute delivery — and it surprised me Baba took so long to react, because quick commerce will take share from traditional e-commerce. A few years ago Meituan delivered iPhones at launches — a wake-up call for JD. The new delivery war kicked off with JD’s initiative around April; JD spent heavily to buy consumers, and Baba joined a month or two later.Money could be better spent elsewhere, but I understand Baba — if they lose relevance in e-commerce, other businesses stop making sense. E-commerce is the core.Despite growing daily volume from 30–40 million to 80 million — sometimes 90 — it’s discouraging Baba hasn’t improved delivery efficiency much. Meituan was already profitable at around 40 million drops a day by carrying multiple deliveries per trip and improving dispatching. It’s sad for investors that many platforms are still loss-making due to subsidies, but Meituan’s underlying efficiency advantage remains. As a consumer, the subsidies are great.Grace Shao (46:29)Why does Meituan have such an advantage in dispatching and logistics compared to Alibaba, which has massive logistics and warehouse footprint?Alan Zhang (47:39)It comes down to the core. Meituan built it through local business development — ditui — integrating merchants into inventory and payment systems. Inventory is kept locally, so Meituan focuses on dispatching and rider movement. Their algorithm can even predict demand and move riders toward hotspots ahead of time.JD invested heavily in centralized logistics hubs and infrastructure — that makes them slower to pivot. Baba used an asset-light model early, working with ZTO, and is more centralized — mostly Hangzhou. Meituan is more decentralized and localized. In quick commerce, doing well in one city doesn’t guarantee another — but once dominant, you can use profit from one pocket to subsidize another. Traditional e-commerce is more centralized.Grace Shao (50:03)That’s a fascinating lens — culture and management style mapping to business model outcomes.Alan Zhang (50:04)And risk. In food delivery, you can’t hold inventory. Meituan works on the assumption you don’t hold inventory. Baba and JD have more of a culture of holding inventory and keeping products in storage longer.Grace Shao (50:27)Closing: biggest prediction for China tech in the next 12–18 months?Alan Zhang (50:51)One for EVs, one for internet. In EVs, OEMs with a pure domestic focus and without an ecosystem will lose relevance in 12–18 months — consumers are making up their minds. In internet, with Tencent hiring Chief AI Scientist Yao Shunyu, we’ll see more AI functionality built into Tencent’s ecosystem.Grace Shao (51:54)What’s one company or subsector global investors are sleeping on?Alan Zhang (51:56)Healthcare. It can be resilient regardless of overall spending. The market is focused on frontier-model spending and ROI, but healthcare companies aren’t budgeting for “latest and greatest” models — they’re looking at applications that improve products and ecosystems. Even if we stopped advancing frontier models for four months, there’s tremendous value to extract from current models.I see a mindset shift among healthcare executives to build AI into products and sell superiority — historically, tech adoption was cost-driven; now it’s revenue-generative. Mindray in Shenzhen, or MicroPort in the Yangtze Delta — great companies. Surgical robots and medical devices are not far behind other systems.Grace Shao (54:15)Final question: what’s one differentiated view you have that’s non-consensus?Alan Zhang (54:37)Instead of focusing only on AGI timelines or capex or cloud consumption, we should think about daily businesses and smaller-scale businesses extracting real value from AI — even financial companies. I’m excited to see new form factors and more AI functions in consumer products.Grace Shao (55:24)So focus on practicality and real use cases — not just headline spending.Alan Zhang (55:33)Absolutely. Look beyond the top three, top five — and don’t go too far down the risk spectrum.Grace Shao (55:39)All right. Thank you so much, Alan. Thanks for your time today.Alan Zhang (55:43)Thank you, Grace. Pleasure to be here.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Z.ai/ Zhipu: one of the first major LLM start-ups to go public. Competition with giants and aims for AGI 29.12.2025 51λ

  In this episode, I sit down with Zixuan Li, who leads the chat API and global partnerships at Z.ai, one of China’s leading LLM labs (one of the four tigers) and now one of the first to head toward an IPO. Z.ai started as THUDM, a Tsinghua data-mining lab best known in open-source circles for GLM and CogVideo, and has since grown into a model-as-a-service platform powering millions of devices and thousands of enterprises in China and beyond.We talk about what it actually means to be an “independent” lab in a market dominated by platform giants like Alibaba, ByteDance, and Tencent, why Z.ai pivoted from SOE-heavy infrastructure projects to a product-led GLM stack, and how they landed on a different business model, and the creation of the GLM Coding Plan, instead of charging by tokens. Zixuan is very candid about pricing (“If Anthropic charges $200, we charge 200 yuan”), the realities of on-prem-first China vs cloud-first West, and what it’s like to race against Minimax and Moonshot with fewer GPUs and less cash.We also zoom out and look at China’s AI talent pipeline (and the meme that the AI race is “Chinese in China vs Chinese in the US”), how he thinks about AGI as self-learning agents that live on your phone, why he’s comfortable being a white-label backbone in the Global South, and where he sees China’s AI landscape in the next 6–12 months. If you want a ground-level view of how a Tsinghua spinout is trying to survive, and maybe win, in the LLM wars, this one’s for you.Newly launched (Dec. 22) GLM 4.7: In today’s world, there’s no shortage of information. Knowledge is abundant, perspectives are everywhere. But true insight doesn’t come from access alone—it comes from differentiated understanding. It’s the ability to piece together scattered signals, cut through the noise and clutter, and form a clear, original perspective on a situation, a trend, a business, or a person. That’s what makes understanding powerful.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.For more information on the podcast series, see here.01:20 – From THUDM to Z.ai: rebrand, Tsinghua roots, and model-as-a-service03:30 – Quiet period & IPO: pride, pressure, and the business challenge of LLMs06:33 – Pivoting from SOEs: infra projects, agentic models, and why strategy followed capability07:25 – Competing with Minimax, Moonshot & DeepSeek: focus, compute, and capital constraints08:34 – Chasing benchmarks vs real-world IQ: math, humanities, and alignment trade-offs11:05 – On-prem vs cloud: why Chinese SOEs still won’t touch APIs13:43 – Zero-retention and trust: can China’s culture around data ever shift?14:07 – Inventing the GLM Coding Plan: subscriptions, stickiness, and “pay by value, not tokens”16:00 – “If Anthropic charges $200, we charge 200 yuan”: pricing strategy and margins and GLM’s open-source flywheel19:41 – Who really pays: sticky indie devs, big tech customers, and bargaining power23:32 – GLM Coding Plan vs Cursor/Qwen/Claude: plans, agents, and avoiding lock-in25:57 – Z.ai’s AGI ladder: AutoGLM, self-learning, and personalized weights27:03 – Independent labs vs platforms in China: speed, resources, and “dirty work”29:34 – Moonshot vs Z.ai: chasing the moon vs being “down to earth”30:53 – Will China’s LLM market consolidate?: 5–10 players, Doubao, and video-generation winners31:44 – Doubao phone & Honor partnership: bargaining power with OEMs34:11 – Beyond China–US: Global South strategy and being a white-label backbone35:29 – Being comfortable as infrastructure: letting others own the brand38:05 – Who joins Z.ai and AI talent: thriving with scarce resources40:07 – Culture, 007 hours, and survival: what it takes to be infrastructure42:33 – Social welfare, AI safety, and cheap tools in India & Indonesia44:38 – How China actually talks about AI safety (or doesn’t)47:29 – Differentiated view: why Zixuan believes you should “enjoy lacking resources”AI-Generated TranscriptGrace Shao:Hey Zixuan, thank you so much for joining us today. Really excited to have you on. Walk us through your journey and what led you to Z.ai to start off with.Zixuan Li:Yeah, so currently I’m the head of Zhipu AI’s chat API services and also head of global partnerships. I collaborate with LMSys Chatbot Arena, OpenRouter, Vercel, these large companies, and ship our products through their platforms.The reason why I joined Zhipu is it’s one of the leading AI labs in China and I can do overseas businesses, because I have a background at MIT’s Schwarzman College of Computing. So that brings my knowledge into real-world practice.Grace:I see. Was there any incentive for you to move back to China versus stay in the US?Zixuan:I think it’s more personal, because my wife’s based in China and she’s used to her work, so there’s no way she can move to the US.Grace:Fair enough.So let’s talk about the company’s mission and origins, because I think it does seem a bit mysterious, especially to people outside of China. From the outside, people know Zhipu, Z.ai as one of the leading Chinese LLMs. But that doesn’t really capture everything you guys do, right?In your recent prospectus, you describe yourself as a MaaS — model-as-a-service — company first. So tell us about that.Zixuan:Okay, so before Zhipu AI, we were called Zhipu or THUDM, because we named ourselves by the AI lab’s name. We originated from Tsinghua University’s data mining group — THUDM. But I think it’s hard to pronounce, and also “Zhipu” is also very hard to pronounce. So this year we bought the Z.ai domain and finally changed our name to Z.ai.When we were called THUDM, we were very famous inside the open-source community because we had a lot of repos, a lot of models under the THUDM name. And we open-sourced not only text models, also CogVideo, CogView, these models. I think they were sold at that time.But with the launch of VEO, Hailuo, and also a lot of current top models, we began to be more focused — basically more focused on text models, visual understanding, and so on. So I think that’s the origination of the lab.But as you said, there’s this terminology called model-as-a-service. From our side, when we compete with large companies like Alibaba and ByteDance, we need to be more focused. They have their inference level, they have their cloud services, but we don’t. So we try to let the model itself provide the service — like the API, or technologies like visual understanding — and try to use the model itself to be the selling point.Grace:I definitely want to double-click on how you position yourself compared to peers — a few of them you just mentioned, whether it’s Minimax and Moonshot, and then you also mentioned the BATs.But to start off with, you’re currently in your quiet period as your prospectus just hit the public. And if successful, you will become one of the first major LLM startups globally to be listed on a stock exchange. How does that feel?Zixuan:I think we are proud of it, but things are very challenging, because it’s really hard to do LLM inference. Both OpenAI and Anthropic have very high revenue, but a lot of loss on their income statements. So we have to figure out how to make money from large language models and also provide cheaper service to the customer.So I think it’s only a starting point for us.Grace:Definitely. I think right now only the big tech companies in many ways are essentially seeing ROI, and the model companies and the model labs themselves are really finding it hard to make a profit.I want to ask you about the branding. You did say you guys changed your company’s name to Z.ai this year, partially because Zhipu is just hard to pronounce. But was that also related to the fact that you guys seem to have made a pivot into really focusing on going global? Z.ai seems to be a lot more non-Chinese-native-speaker friendly, right? So is that the push right now?Zixuan:I think that played an important role, because we have observed the success of DeepSeek, Qwen — they got famous globally and Chinese people will think that they are the “SOTA” in the domain and their models are the best. They are recognized by NVIDIA and other large company CEOs. So I think that’s one factor.But the other factor is when we changed the name to Z.ai, the dot also plays an important role. We want people to enter that URL into their browser and try to visit our website. Yeah, two factors.Grace:And tell me about your origin story, actually. You mentioned earlier you started off from the Tsinghua data mining group. Maybe provide some context to people outside of China. What does Tsinghua represent? I mean, it’s an institution, it’s a university, but why are so many of these LLM companies or even deep tech companies coming out of Tsinghua right now?Zixuan:I think it’s kind of a combination of Stanford and MIT. So talents are everywhere and there’s a lot of funding from internally and also externally. And also people are chasing the highest IQ there. So it will be very natural to pursue AI in Tsinghua University.Grace:So I have a question on that, because a lot of tech companies, even the previous generation internet companies that came out of Tsinghua, had some kind of connection with Beijing city. And my understanding is Zhipu’s original business model was also very focused on SOEs and local government work, both in China and even across Southeast Asia.Before the more recent pivot leaning into tools and APIs, what were the reasons for the pivot from the heavy AI infrastructure focus and SOE projects to a much more product-led tools and API strategy?Zixuan:I think it depends on the capabilities of the model, because nowadays the model can perform agentic tasks, use tools, use coding to perform tasks. But before that, we could only do customer service, data processing — these “dirty work.” I think it’s better for SOEs or other scenarios.But with the change of Cloud Code, GLM 4.5, these agentic stuff, people can really use the model in other areas like Manna, Gainsburg, Lobe. So I think it’s not only our strategy, but also the capabilities of the model have changed a lot.Grace:Yeah, and I think to put you on the spot, where do you see yourself compared to your peers — like the DeepSeeks and the Minimaxes and the Moonshots of the world?Zixuan:I think compared to Minimax and Moonshot, we are close competitors. We are startups, but DeepSeek is like another kind of enterprise because they have Qwen. So I think they’re very unique, and also ByteDance, Alibaba, they’re sitting at the same table. So they’re from large enterprises.We are all chasing somehow the same direction, but we lack compute, we lack money compared to these giant enterprises. So we need to stay very focused.Like Moonshot, they focus on the Kimi K2 series. They only release Kimi K2, another K2 and K2 Thinking this year. And also Minimax — they’ve become more focused and they kind of shift away from multimodal to text models. I think it will be very fierce. The competition will be very fierce in the coming months.Grace:And for yourself, when you say you’re chasing the same direction, what does that direction look like in layman’s language?Zixuan:In layman’s language, I think… more practical. Because the reason why we do coding and agentic is that we see people using it. We see people using Codex, Manna, Claude Code. So that represents high token usage.And also we are chasing AGI, or the IQ. So we want the model to solve very hard math problems, to memorize a lot of hard stuff. As you can see from a lot of benchmarks like ARC-AGI, HLE, we’re also chasing in that way.So we balance the two: figure out how to balance the performance on benchmarks and in real-world development.Grace:I actually have a question on that that’s a bit off-track from the business strategy side of things, but I wonder how you view this.So you’re saying you’re chasing benchmarks on math problems, IQ, advanced physics, etc. But what about the humanity side of things? I think people are still questioning whether AI can be used to replace humans in a more humanities-focused industry or sector.Zixuan:So that’s a very big issue. But for now, I think it’s still not there yet, because we see hallucinations happen inside the model and instruction following is not very good.So we test the status of that harm and try to assess what we can do with this model and try to synthesize a lot of data to make it more aligned to human judgment or other things. I studied alignment at MIT. I know a lot of stuff, but when I came back to China, I thought we were not there yet.So capabilities, I think, are still more important than alignment at this stage. But we need to focus on the future and try to prevent something really bad happening. I’ve learned a lot of news like suicide or emotional feelings, depression. But somehow I think it’s still not that harmful yet.We try to incorporate as much human judgment or human alignment into the model as we can. But as I said, it’s kind of a balance between different aspects.Grace:Yeah, it’s always a balance between setting up the guardrails and actually still allowing the technology and innovation to continue, right?I want to reshift the focus back on business model, pricing, deployment. Reading the prospectus, what stood out to me was how much you support both on-prem and cloud.What are the main product lines today of Z.ai or Zhipu, and how do you map those onto on-prem versus cloud deployments in terms of how customers actually adopt GLM? Because I do believe I read that in China there’s a very different preference. In China, it seems like more people prefer on-prem, right? Whereas in the US it’s more cloud — or did I understand that incorrectly? Please explain.Zixuan:Yeah, I think you have a very good understanding of the current status, because large SOEs, large enterprises in China, prefer on-premise or more private deployment. So it’s hard to do API services with them.But currently a lot of tech companies accept API services. So we collaborate with nine out of ten of the largest Chinese tech firms or social media firms with our API services. So it depends on their needs.We try to sell API, but actually some people have privacy concerns. They have policies not accepting API services. They don’t want any data to go away from their servers. So basically it depends on the users’ needs.Grace:This is actually kind of a reflection of what happened during the SaaS era too, right? Chinese SOEs and big companies would rather build their own app — maybe not even be as good — but they just don’t want to give their data out to anyone and have that potential security risk, right?So do you think that will change in terms of company culture as we see AI continue to develop, or do you think that will continue to be the trend in China — that this would be the differentiating point between the Chinese market and maybe the Western markets?Zixuan:I think it will continue to be the trend. As you said, we had that pattern in the era of SaaS. And when we go to the AI era, nothing changed.But somehow, we can figure out a way to balance, because there is more “private host on cloud” service. And we’re trying to store user data in a more secure way, with a zero-data-retention policy. That will mitigate the risk and the issues and try to let them feel more comfortable with it.Grace:I see.A lot of Western developer tool companies now go pure usage-based, but you guys also have a GLM Coding Plan — basically for developers with very low entry points. Why did you choose a subscription approach versus going with other pricing models? I guess this part, I just want to understand how you guys are making money right now, especially as you’ve just had your prospectus go public.Zixuan:Yes, I think we are the company that invented this coding-plan business model, because we found out that API users are not sticky. One day they use Claude, they can switch to Gemini or GPT another day. It’s the same with Chinese models.So we remembered: why do we pay for subscriptions — Spotify or YouTube service? Maybe we just listen once during the whole month, but we don’t regret it, right? So we don’t want our users to pay by tokens. We want them to pay by value or by the product itself.So if they just use it once or twice within a month, I think it’s totally fine. If they want to subscribe the other month, we’ll try to provide better service. We have GLM 4.5, GLM 4.6, GLM 4.7, trying to ship better models. But if they decide to quit, I think it’s still good for us because they paid for one month, not just several tokens.So the users may be very sticky here, and we have our branding — not only the model, not only GLM, but also the subscription, GLM Coding Plan. So when we use Cursor…Grace:But if they were to use it a lot, would it be loss-making for you guys then?Zixuan:I think it’s still an issue for Claude Code and also Codex. You have to balance the rate limit and also the service level. So for us, we are very generous, but we’re trying to operate globally, because that will make our traffic more stable — not receiving very high demand at one time and no demand during the nighttime.Grace:I see.I know that you’ve been quite active in a lot of podcasts recently. You were on ChinaTalk, you were on Steven Hsu’s. And one of the lines you said, I think it was on ChinaTalk, you said, “If Anthropic charges $200, we charge you 100 yuan.” I thought it was quite funny. It was very memorable.So how did you make that kind of decision, and how does it work in practice? Does that mean you have a long-term structural advantage, or does that mean you charge less and therefore have smaller margins?Zixuan:I think we serve different customer needs. For example, someone sells Rolls-Royce to people, but we sell Benz to people. Both are good cars, but Rolls-Royce charges way more than Benz. The performance, I think, is very close.But like I said, Anthropic deserves that premium. But by selling Benz, we can still earn a lot of money. Maybe the profit margin is very thin currently, but we can lower the inference cost. We can change our infrastructure to make it more profitable. So it’s a long-term strategy, not focused on the current cost structure.We’re trying to make people more sticky to the brand, more sticky to the service. I think it’s essential at this time.Grace:Essentially, you’re saying the utility purpose of having a car — getting from point A to point B — is the same, but maybe you’re selling a Toyota then or a Honda, right? Not even Mercedes, which still charges a pretty premium margin.I remember in the same interview, you were kind of challenged, saying: look, you only really take up about 5–6% market share in China for general-purpose models. But you said, “Wait, 5% is enough.” What exactly are you thinking when you say 5% is enough?You serve — I think from public disclosures — 123 large enterprise clients on-premise deployments, plus around 5,500 customers using cloud services. How does that actually stack up to your peers? Because it doesn’t look like huge numbers, to be honest. And that already is 5–6% of China’s market share.Zixuan:I believe that the 5% refers to the percentage of all the GLM services.Grace:Yes, sorry, GLM.Zixuan:GLM services, because we open-source our models. And it’s hard to get revenue when you open-source your model because you have to compete on speed and stability.But I think our model is good enough. Maybe it’s not like Toyota — it’s kind of a Benz. And we let more people adopt GLM, like what Qwen did in the past. They open-sourced their reflection models and more people tried out Qwen. They got famous, so people believed they got better service from Alibaba.It’s the same underlying methodology from our side. So if GLM gets really famous, even 5% is enough for us. But if it’s not famous, 5% is totally not okay. We’re trying to make our model more influential, like DeepSeek, like Qwen.Grace:I see. I do want to go into GLM and your tools later as well. But one last question on the business side of things. We kind of touched on this: you said a lot of your customers are the big tech companies, but in the beginning, they were the SOEs, right?So right now, is there any pattern you’re seeing in terms of who becomes the most valuable users and who becomes the most sticky users and who are actually willing to pay the big bucks for your product or for your service?Zixuan:So from my department, I think two types of customers. One is individual developers, because we have the GLM Coding Plan. Someone bought a yearly max plan. A lot of users bought yearly plans. They are very sticky.And the other type is large tech companies, because we are still leading the open-source models. So we have bargaining power. Maybe they want to shift away from our model and choose other models, but we keep evolving from 4.5 to 4.6 and 4.7. Every time they try to change the model, they find that we can ship better models.So these customers are very sticky. And they care more about performance because we are leading in performance. They care less about cost or relationship.Grace:Nice.Let’s actually double-click on GLM. You mentioned GLM 4.5 and 4.6. They’ve been positioned as highly competitive on coding and reasoning, and you’ve often been the highest-ranked Chinese model on public leaderboards.When you compare the GLM series to US and Chinese peers, what dimensions matter most to you beyond the leaderboard scores right now? And where do you think GLM actually genuinely stands out compared to other peers, whether it’s American ones or Chinese peers?Zixuan:I think real-world development, real-world practices, and general chat — these real practices — are more important than benchmarks. And in terms of real-world experience, we are tier two, because I believe Anthropic, DeepMind, and OpenAI have better user experience compared to us.But I think we are enough compared to other open-source models, because we understand user needs. We have better quality in data — pre-training data and post-training data — and we’ve figured out ways to synthesize agentic tool-use trajectories and very hard problems. That makes us stand out in solving these really tough problems.Because when you look at the benchmarks, they are not for real-world practices. Some are very tough, but it doesn’t mean they stand for human practices. Because we have a lot of customers… yeah.Grace:Yeah, I’m going to challenge you on that actually. What about the Alibabas of the world? Because when I speak to Alibaba or Tencent, they also say their biggest differentiating point is real use-case data. And frankly, they have all the existing touch points with their users, whether it’s getting data through helping with businesses, enabling businesses, or consumer use. They probably have the best data, right? So how do you compete with that?Zixuan:I think that’s their advantage in 2024, but not 2025. Because in 2025, most of the high-quality data we need, we have never met in real-world use cases.When you want to create a slide, you first do search and then come back and do another round of thinking, and then choose a design tool or something like that. Nobody interacts with Alibaba’s product like that. So you have to fully understand Cursor, Claude Code, Manna — how these tools interact with people.So ByteDance and Alibaba’s customer data cannot play a role in today’s agentic era. We have understanding of maybe Claude Code or Codex — we try to understand how a top-performing agent manipulates tools and how our model can be integrated in that system.Grace:I was actually going to ask you about the GLM Coding Plan. So for context for listeners, it’s essentially their tool, like a Cursor tool.So how does the GLM Coding Plan actually compare with Cursor or Alibaba’s coder, as you mentioned, or Claude, in terms of coding experience? For a developer who already knows these tools, how would you explain the distinction — or, you can be frank, is it mainly a pricing advantage here?Zixuan:Okay, so I want to compare Cursor with GLM Coding Plan, not the model. Within Cursor, you have one coding agent and you can switch between different models. But with GLM Coding Plan, you first select the model and then you can switch between different tools.You can integrate GLM into Claude Code, Kimi Code. You can even use GLM in Cursor with GLM Coding Plan. That made our product or model widely accepted or widely integrated into these systems — not just for Claude Code, but also it can be integrated into Cursor or Kimi Code.We understand different coding agents and try to synthesize data that best fits these coding agents’ needs. And there’s no lock-in for our users.Grace:So your GLM Coding Plan is not only your proprietary model, right? You actually are open to multimodal?Zixuan:Yes, it’s a model. GLM Coding Plan is called a plan, not an agent, not something like Claude Code. You subscribe to an API, you’re not subscribing to a product. You use that API maybe in Claude Code, maybe in Kimi Code. So you can choose the mode.Grace:Yeah. Okay.Okay, thanks for explaining that to me. That’s helpful, I was getting a bit confused there.Now I wanted to ask: in your prospectus, you laid out five stages of progression into AGI. We talked about your vision of AGI earlier. You said it’s about real-life implications, real-life practicality, usage of AI.When you look at where you guys are at right now, what does crossing the next stage look like in terms of concrete capabilities or products or tools? Or maybe a more straightforward way of asking this is: what should we be expecting from you guys in 2026 to help you progress on your so-called AGI pursuit?Zixuan:Maybe self-learning. Because currently when we do reinforcement learning, we synthesize all the data, we prepare the data beforehand, but the weights of the model won’t change during the interaction.For example, we have this AutoGLM. It’s a model that can be deployed on your phone and can manipulate different apps for you. It can order food or order an Uber for you, but it’s the same model for everyone.To chase AGI, we might have AutoGLM for everyone. When you interact with the model, the weights of the model may change. Currently, we have a memory engineering package that’s more on the engineering side — handling this memory stuff.But for AGI, it needs to be very personalized. Every model needs to be personalized. The model learns from the environment, from the interaction. We also call it on-policy reinforcement learning.Grace:I see.Let’s take a step back and look at China’s overall LLM landscape and competition. You kind of alluded to this earlier — you guys are in the same pool as the Minimaxes and Moonshots of the world. Then there are the big techs like Alibaba, ByteDance, Baidu, Tencent, even Huawei these days, right? There’s so many. Everyone’s producing their own LLMs now.From inside the ecosystem, what do you see as the structural differences between independent labs versus the big tech platforms — in terms of commercialization of their models as well as their incentives and objectives in the coming year or two?Zixuan:Strategy and objectives. Because we lack resources, we need to be very focused. And when we are very focused, we need to move very fast.For talents, our team is very small. I lead a team…Grace:It’s not that small — a couple hundred, right? You guys have like 800 people now?Zixuan:But for every team, there are just a bunch of people. We have sales, we have product solution, but for the product team, product solution, or training team, sometimes you need to be very lean. You don’t have to hire a lot of people, because they chase different directions.Sometimes you have to hire people that can do “dirty work.” Maybe one person is enough to do all the training on this side, and you have a bunch of people preparing data or understanding customer needs for you.Like I said, you have to understand Claude Code, you have to understand these coding agents. So there will be people studying all the products, looking inside these products to see why they are performing so well.But for large enterprises, they can hire a lot of researchers. They have enough resources to do a lot of experiments. They have compute, so they worry less. Maybe they can find some scientific breakthrough from those experiments.But in terms of model performance, I think our competitive advantage is we are closer to users and customers, because we move faster together with our users.Grace:So that’s how you position the startups versus incumbents. But what about just within the startups yourselves? How do you differentiate yourselves between one another?Zixuan:I think compared to Moonshot — because we both originated from Tsinghua University, we know each other pretty well — I think we are more down to earth. We are the ones that care more about real-world usage or practices.Moonshot is kind of… they have this “AGI plan,” chasing the moon or landing on the moon, and they have more imagination on the surface. We’re also chasing AGI, but when we train the model, we care more about real-world practice and usage.Grace:You’re taking a more pragmatic approach. And they’re definitely, I think, a very eccentric bunch, right? Even the name — how it came about — was quite interesting.So do you think eventually in the Chinese LLM space it’s going to be winner-takes-most? Maybe not winner-takes-all, but winner-takes-most? Or is it going to be able to support multiple strong players?Because there’s been rumors about consolidation for a while. There are quite a few players for how big the market is, and like you said, it’s extremely capital intensive. Not everyone has this much money to keep burning through it. So where do you see the direction of this fragmented landscape right now?Zixuan:I think the market is enough to include 5 to 10 players. I think it’s enough. And like I said, the large enterprises only accept on-premise deployments, so there’s no way a winner can take it all, because there are thousands of large enterprises. You don’t have the team to deploy models for every single enterprise.But in terms of applications, maybe Doubao will take more than half of the consumer side. And also for video generation, there will be a winner. But I think the market is still very large to have all these players, and they will compete for a long time. I can guarantee that they will compete for a long time.Grace:What do you think of the Doubao phone situation? This is completely random. This is not relevant to our LLM conversation, but I’m quite curious to hear your thoughts on it, because I think it’s making a lot of noise outside of China. People are quite curious to see where that will lead to.Zixuan:So we are the first company to launch this phone use agent. But I think the issue is bargaining power. We also collaborate with a phone company, and instead of using something like a “GLM phone,” we finally used their name. Their phone, powered by our model.Grace:Which phone is this?Zixuan:Rongyao.Grace:Okay — Honor. I think it’s called Honor, yes.Interesting. You know what? I really haven’t heard about it, but I should look into it. Is it actually already available to the mass market or no?Zixuan:I think the phone was launched last year, not this year.Grace:Okay, super interesting. I’ll look into it.Zixuan:Yeah, so a lot of phones at that time were powered by AutoGLM’s capabilities. But we don’t have the same bargaining power as ByteDance, so we cannot name the phone by our name. We just power their scenarios.So it’s about bargaining power, I think. Because like there’s the ByteDance vs Tencent issue, also with WeChat — it really depends on how you split the revenue, the value, how you make sure that you won’t influence other people’s business.So finally, you have a line: maybe this app will collaborate with you, and that app rejects your endpoint.Grace:Yeah. For context for you listeners, WeChat rejected Doubao phone’s direct access, and there was a huge headline war on this like two weeks ago.Okay, I want to pivot a little bit. Right now there’s a lot of focus on the China–US lens. And you yourself spent time in China and the US as well.But I did notice in the beginning days of Zhipu you guys were actually really focused on the so-called Global South — for lack of better words — Southeast Asia, Latin America, maybe even Africa. Is that still a strategy you guys are pursuing? Looking to sell or actually embrace markets that go beyond just China and the US?Zixuan:Yes, definitely. Because I think in GLM 2, GLM 3, we only had Chinese and English capabilities, but now we have more than 100 languages. So that can support us going beyond English-speaking countries. Maybe in Brazil, maybe in Malaysia, we have opportunities to showcase our model or showcase our product solutions to people and finally compete with those large enterprises.But I think things are really different in those countries, because they also want their data as private as possible. They accept on-premise, and maybe they want white label — they fine-tune the model and they want to ship it to their citizens under their name, not GLM or Zhipu’s name. So we have to meet their needs and see what we can offer.Doing business in the US, I think it’s much simpler because you have this API, you have products, you can do a coding agent, you can earn money. But when you do business in other countries, you have to go really deep, twist a lot of things, and try to make it happen.Grace:It’s also interesting — I think you touched on something. You’re quite comfortable being that white-label provider, versus I think a lot of other companies, whether it’s ego or belief, are not as comfortable. They definitely want their name on it.So it seems like you guys are actually the backbone supporting a lot of technology or clients without really having your name attached to it.I want to ask you about talent. This is a question we touched on in the beginning — you said yourself you came back to China for personal reasons, because your wife is in China. But I assume that’s not the case for everyone.There’s this interesting and funny joke going around saying right now in the AI war or AI race, it’s really between the Chinese in China and the Chinese in the US. It’s just funny — there does seem to be a high percentage of ethnic Chinese or Chinese nationals or Chinese-naturalized Americans or ABCs. If we’re being non-PC, people who look Chinese in the field.Why is that? I don’t understand. Did Chinese people just get a tip-off saying AI is gonna be really big early on and they went into this field earlier, or what happened?Zixuan:I think I cannot explain it, because doing math problems is simple for us. I’m not sure why other people won’t pursue this business.Because when I did internships and research at MIT, I saw a lot of talented people beyond Chinese — they’re still talented. They finally went to Anthropic, OpenAI. But somehow people only care about Chinese because they are co-launching products with Sam Altman or Elon Musk.I think people overrated the influence of Chinese people in the large language model area, because still there are a lot of enterprises not relying on Chinese.Grace:It’s quite funny — it’s kind of like the last generation, where every Chinese student in the US is either studying to be a lawyer or a banker, and now everyone switched over.Actually on a more serious note, how does the talent competition play out then? Do you see yourself at Zhipu having to really convince people to join you compared to a US peer?Or do you think there’s certain tendencies for certain researchers that would prefer to work for a Chinese lab or return to China? How do you see that play out?Zixuan:I think we finally choose the people that best match our environment. Like I said, we lack resources, but some people really enjoy the lack of resources — like me. Because I think it’s good to have a small team competing with a very large team, and you have better enjoyment when you conquer a puzzle or problem, or you finally win at the end.So people who enjoy this feeling, we try to hire them. And like I said, we want to move really fast. We want people — both the product team and the training team — to understand the user scenarios, to understand the data itself, not just theory or the algorithm.So we try to find those people, and they will finally choose us because they don’t care about compute or resources, or they find it too toxic competing with other teams doing the same experiments and the same thing. Because that happens a lot in large enterprises — a lot of teams doing the same thing.Grace:Yeah, for sure. I think even when I speak to the BATs in China, there’s so much internal competition that drives people crazy. It’s internal politics that drives people crazy. But that also becomes an incentive for people to really push.On that note, you guys are about what, 800 to 1,000 people altogether, roughly around 100 to 200 in R&D — something around that rough figure. It’s essentially not really a startup company anymore — it’s just small compared to how big the big tech incumbents are.So at this size, and as you guys head into becoming a publicly listed company, do you see the culture changing? And what are the ways you keep your researchers, scientists, and engineers motivated? Are we seeing crazy salary numbers as well, like the ones coming out of Meta? How do you keep people motivated?Zixuan:I think we are more lean, more entrepreneurial. Especially in our team, because I only slept 50 minutes for the past 24 hours. So we want to move really fast, faster than everyone else. Yeah… beyond that.Grace:You’re going beyond 996. This is not 996, this is 007.Zixuan:Because the competition is really fierce. Moonshot, Minimax — they’re doing an excellent job. And we also have DeepSeek, Qwen — not to mention the frontier AI labs in the United States. So we have to keep pushing. I think there’s no other choice.Because when we try to do AI, we want to survive. Frankly speaking, survival is a very high standard for the tech industry. When we look at operating systems: Windows, macOS, Linux — I think that’s enough. And when we look at phones — only Android and iOS.So the competition must be fierce when you want to be the infrastructure for the industry.Grace:Yeah, I agree on that. Okay, well, I hope you get some rest soon after this call. I really appreciate you jumping on the call after 50 minutes of sleep today.Looking at your long-term vision and where you guys are headed now, especially with an imminent IPO: in your public materials, you talk about AGI integration with the physical world and social welfare as a long-term vision. I think this is something not many AI companies frankly are really thinking about.Even within our conversation, you’ve talked about the balance between tech acceleration and actually putting up safety guardrails, essentially to prevent more sad, tragic happenings caused by AI psychosis, etc.When we look at this, how do you personally reconcile the social-welfare North Star with the commercial realities and the pressure you just talked about? Where are the areas where you guys are frankly more okay to let go a little bit for business gains? What areas are definitely your red lines that you cannot cross, where you really want to hold people accountable and ensure there are no AI-caused tragedies?Zixuan:Yeah, I want to answer this by giving an example. We have this GLM Coding Plan — it’s very cheap, three dollars a month. A lot of people in India, Indonesia, or even in the United States use GLM Coding Plan to do their side projects or even their startup.I just talked to a person today. He’s doing a startup that uses GLM Coding Plan to write a program that can collect recyclable bottles. They scan the bottle and recognize it, try to differentiate trash from recyclable products, and make it a real business. So we truly use AI to empower these businesses. You can see there is a lot of social welfare behind this.People just use the coding, but you can use coding to do a lot of stuff. We provide the service, but we let people decide whether they try to contribute more or only care for themselves. So I think it’s a starting point for us.With more powerful products — maybe next year — we can empower larger scenarios. Maybe we can empower robots.Grace:And in terms of this topic, I think in the US there’s a very dominant voice and discussion about the potential risk of AI or the negative impact it might have on society. AI Proem and Differentiated Understanding, frankly, very much focus on the business strategy of technology. So a lot of my guests and myself, we focus on capital deployment and feasible business models.But I do want to ask: you’re plugged in, you are in China, in the LLM space, in the AI space. Is there a discussion about AI safety, or are people really just quite focused on acceleration and pragmatic deployment and diffusion?Zixuan:I think compared to the United States, not that much. It’s more pragmatic. But that’s still on people’s minds, because AI safety is still an issue for us.We can see the ceiling, the threshold of all the current capabilities, and understand what’s the top priority for our model or our scenario, and try to fix those before going to the next step. But we always keep the security and safety issue in our head. And when that day finally comes, we can be fully prepared.I’m engaged in a lot of these conversations in the US and I’m also part of Concordia AI. It’s an organization focused on AI safety. I’m part of it in Beijing. But when I left that company, I saw them — and for anyone talking about this — it’s not because people don’t care. We can train a model with better capabilities and also a safer system.So there’s no trade-off at the current stage because we don’t have to balance performance with safety concerns. We can improve them at the same time.Grace:I see, it’s more like taking a mindful approach.I want to end on two quick questions. One is: usually people ask, “Where do you see yourself in the next five years?” Right. But I think for AI we can’t really ask that right now — no one will know what five years looks like.But for our listeners: where do you think China’s AI space will look like in, let’s say, six to twelve months? Where do you think the focus will be, or the potential breakthrough?Zixuan:Potential breakthrough may be integration with the physical world. When we see a lot of robotics companies and we see a lot of smart glasses, people are shifting focus from AI companies to these, we call it broader intelligence companies. So that might be a shift.And also DeepMind — I think they’re doing the same path. When you look at Gemini, it’s not just a large language model but also it can perform world knowledge or integrate with real-world use cases.When we look at Gemini 3 Pro use cases, someone is controlling the camera or trying to integrate with the computer. So there are a lot of things we can do with large language models.Grace:Okay, I think the last question I have for you is a question I ask every single guest, which is: what is one differentiated view you have — a non-consensus view? It could be about anything: about the industry, about how you see the world.Zixuan:I think for me, I’ll just share my thought: you should enjoy lacking resources — lacking people, lacking everything. In the AI world, that pushes you to the boundary. That pushes DeepSeek to change their architecture, to really do something innovative.For me, I don’t train models, but I build products and do marketing. I had this GLM Coding Plan thought because we don’t have very loyal customers. When they use API, they try to shift from GLM to someone else and then come back one day. So I noticed these difficulties. That’s what we aim for: to try to solve these really tough difficulties.Grace:Yeah, I think to your point — when you lack resources, it also means you have the agility and the flexibility to change things, because there’s no bureaucracies, there’s no chain of command, and it’s much faster.I appreciate that in itself too. I was talking to a friend about that recently as well. Since I left big tech and left traditional media to do this myself, you have so much more flexibility, and sometimes you’re upset that you don’t have the access or the resources you used to have. But it does help you build faster and connect with your community faster.Thank you so much anyway, Zixuan. I really, really appreciate your time. Please get some sleep after this.Zixuan:Thank you too. Yeah, I truly agree that you have your competitive advantage, because those large media companies — their journalists won’t reach out to me. So it’s my honor being here, but also a good opportunity for you to understand the Chinese market.Grace:Yeah, for sure. And I really appreciate you giving me your insights during this time. And for all the other Chinese AI labs out there, if you’re listening to this, please reach out. I would love to have a conversation. Thanks again.Zixuan:Yeah, thanks.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  What the U.S. Misreads About China’s Tech Rise with Kyle Chan 23.12.2025 52λ

  In this episode, I sit down with Kyle Chan (Brookings Institution) to unpack the thinking behind his provocative New York Times op-ed, “In the Future, China Will Be Dominant, the U.S. Will Be Irrelevant.” We start with the DeepSeek moment and why it surprised the West, why it didn’t surprise many China-watchers, and why Kyle sees it as only “the tip of the iceberg.”From there, we zoom out into the bigger story: China’s rise isn’t just one breakthrough model or one champion company. It’s a system of interlocking capabilities: EVs, batteries, renewables, industrial automation, robotics, and AI, advancing in parallel and reinforcing each other through spillovers, supply chains, and fast-moving “Swiss Army Knife companies” like Xiaomi and Huawei.We also dig into what people often get wrong about China’s state role: not pure top-down command, but a mix of industrial policy + private-sector experimentation, including practical mechanisms like compute vouchers and local-government support. Finally, we cover India’s trajectory, geopolitical constraints, and Kyle’s “hedges”—scenarios in which today’s narratives (in both China and the U.S.) could still break in unexpected directions.Relevant links: https://www.brookings.edu/people/kyle-chan/In today’s world, there’s no shortage of information. Knowledge is abundant, perspectives are everywhere. But true insight doesn’t come from access alone—it comes from differentiated understanding. It’s the ability to piece together scattered signals, cut through the noise and clutter, and form a clear, original perspective on a situation, a trend, a business, or a person. That’s what makes understanding powerful.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.For more information on the podcast series, see here.00:00 — The NYT op-ed + the DeepSeek catalyst: why Kyle wrote the piece, what he wanted to correct, and why DeepSeek was a wake-up call (“tip of the iceberg”).06:53 — Kyle’s origin story: infrastructure obsession (high-speed rail) → the path into tech & industrial policy.12:31 — China’s “electric tech stack” + spillovers: EVs, batteries, renewables, robotics, AI moving in parallel—and why “Swiss Army Knife” firms (Xiaomi/Huawei) can leap across categories.19:12 — Why autonomy pairs with EVs: the technical and architectural reasons autonomous systems “almost always” sit on EV platforms.24:01 — China AI ecosystem in practice: startups + hyperscalers + policy “tailwinds” (compute vouchers, industrial parks, local government support) and how that differs from the U.S. model.29:46 — China’s development playbook vs others + the India comparison: proactive bottleneck-solving (“ground game”), plus India’s tailwinds and constraints over the next decade.41:00 — The hedges + the wrap: what could derail or reshape the trajectory (trade backlash, geopolitics, bubble risk, robotics paths), and Kyle’s non-consensus take on policy intervention.AI-generated transcriptGrace Shao (00:00)Hi, Kyle. Thank you so much for joining us today. Kyle, I want to start with your recent New York Times op-ed, which had a pretty provocative headline. It’s called, In the Future, China Will Be Dominant, the U.S. Will Be Irrelevant. When I saw that, I was like, whoa, this guy, someone’s going to go get blood now. How did that piece actually come about? What was the main, I guess, objective or goal out of that piece?Kyle Chan (00:16)Yeah, yeah. Thanks for asking about that piece. Yeah, that piece, it got quite a reaction. I was surprised. And there’s been a number of pieces I feel like now—sort of, it’s almost become a genre of like all the things that China’s doing, all the things that the US is doing, the sort of divergent trajectories of the two countries, especially on technology. And for me, one thing I really wanted to focus on was—So we had the big DeepSeek moment earlier in the year, and that really got people to wake up and take notice of what’s happening in China and China’s tech development in a way that really, I mean, I can’t remember the last time that something like that happened. And so that was quite a big wake up call. But as someone following a number of different sectors in China for a while, I was like, this is just the tip of the iceberg.I mean, first off, within AI itself, DeepSeek—I think it was kind of funny—was surprising for a lot of people who follow actually China’s AI industry quite closely, because I think we might have been expecting some of the bigger tech companies to have made a bigger splash, but DeepSeek seemed a little bit out of left field. But within AI in China in general, there is so much talent, so much engineering talent, a vibrant developer community, top-notch researchers.So like when you look at, say, who is accepted, whose papers are accepted for NeurIPS, one of the top AI conferences, right? It’s many, many, many names from Peking University or Tsinghua University or Zhejiang University. And so if you follow that space for a while, DeepSeek was not so surprising. Maybe it was surprising that it was DeepSeek itself, but that China could produce a world-class AI model on par or nearly on par with some of the best in the US—that was maybe not so surprising.And then it’s not just AI. The other thing is like when you follow EVs or when you follow batteries, or if you follow anything related to clean tech—solar, wind, hydrogen fuel cells. If you follow robotics, anything related to industrial automation, industrial robotics, also self-driving cars, smart driving systems. I mean, the list goes on and on for all these different areas.And then it gets even down to sort of like the basics, right? So like some of these traditional industries where you just see this like classic China chart. Call it like the classic China chart where it’s like the share of global manufacturing for shipbuilding, say, or steel—is like, at first you see like China is growing and then soon it’s like eclipsing the rest of world combined.So to me, it was this bigger story that I really want to highlight: not just DeepSeek and not just AI China, but more broadly speaking, what is this bigger trend and why should we care? How is this going to shape not just Chinese society, but the rest of world?Grace Shao (03:11)I think to your point, DeepSeek was very secretive, yet it wasn’t like it’s within the AI industry in China—people were already noticing it and people were talking about, I think maybe six months before even they came out with their first R1 and then V1. But I think to your point, yeah, it was a shot to the West because it was like, wow, we always knew that China had strong industrial capacities, right? Like you said, like we had the manufacturing capabilities, the factories and whatnot, the hardware capabilities.They didn’t expect something like a software to come out of China that was almost on par with what they could produce in the West. I guess my question for you next is then to highlight that—what was your goal really? What was your real message that brought you public? Why did you publish an op-ed on the New York Times?Kyle Chan (03:55)Yeah, so part of it was to kind of point to the underlying drivers for what was happening because I also wanted to kind of correct this image of China, not only in terms of like China’s tech development, but also what really was responsible for some of that.So like the image I want to correct was basically this very old notion of China making, you know, low value added commodities like household goods, basic consumer electronics maybe—stuff that maybe is good for economic growth, but isn’t so impressive technologically and doesn’t really challenge, say, the US or Europe or other industry incumbents in these areas.I want to first point out that, yeah, this is different China now. And this process has been unfolding for a long time, actually. So I was trying to highlight some of the efforts that the government was trying to do to help accelerate not just industrialization, but innovation itself.This idea that it’s still so deeply controversial in the US—the idea that the government might have a positive role to play in supporting private sector development, supporting cutting edge technology—I think that that is still something that’s debated very hotly in the United States. And I wanted to point out how China has been able to use—not successfully every time, and there’s definitely issues along the way, but overall, quite effectively—it has been able to use industrial policy to really move the needle and support its industries and its private sector.And so this combination too of like, it’s not just one or the other. It wasn’t just sort of all top-down state driven and it wasn’t just all sort of bottom-up private markets. It was this interesting combination that has produced, I think, these sort of like world beating industries.And I think the lesson—a big part of the piece was about the US side of it and what lessons we might take away and how the US might need to step up its game. I don’t know if this competitive framing is the right one, but in general, a realization that, okay, there’s a lot happening. This assumption that China would always be the center of low-cost manufacturing and the United States would be the center of high-tech R&D, innovation, Silicon Valley—that the picture was much, much blurrier than that. So that was sort of like my overarching goal.Grace Shao (06:27)I definitely want to double click on the part where you talk about how the state and the private sector actually work together. And we can talk about that later. But I want to get a sense on what kind of feedback or pushback or even maybe criticism did you get from that piece?Then furthermore, I want to get understanding: how did you get involved in all of this? You’re in the US, right? You’re in New York, right? How did you get into studying China’s industrial policy? Tell us about your background.Kyle Chan (06:53)Yeah, yeah. Yeah, I mean, I just recently joined Brookings based in DC, which is a DC-based think tank that has a really great China center that does outstanding research on policy issues related to China, US-China relations. And my focus now is on China’s tech and industrial policy.But getting to this point has been like an interesting journey. So originally actually—I mean to go all the way back, I don’t know how far you want to go back—but like my family is actually from Hong Kong originally. And so I was born in California.And my parents—it was sort of that generation where my parents really wanted me to learn Mandarin. They’re like, that’s gonna be the useful language. And it turned out to be very useful.But also growing in California, I took cars everywhere. It was like a very, very much like a private transportation kind of city. And it was like a revelation to me—I mean, this sounds so ridiculous to anyone who’s grown up in a city with good public transportation—but it was like a revelation to me to later live in places like Chicago or even San Francisco and then later on Beijing and Delhi and Berlin, to be in places with like functioning subway systems and functioning public transportation.So I got really interested in infrastructure, actually, not necessarily industrial policy. That kind of came later, but infrastructure. And here there is really sort of like a very strong role for the government to play in coordinating, if not actually building and maintaining infrastructure, whether you’re talking about roads, highways, bridges, railways, subways, electricity grids.And I just found it really interesting then later on traveling to China—how this seemed to be like completely different there. I mean, and I remember at the time really being amazed by the high speed rail system there.And China didn’t have a bullet train system for most of its, most of the existence of its railway system until basically starting in the 2000s. They started to take seriously this idea of like, okay, maybe we can like really roll out a nationwide bullet train system. And they did a lot of R&D.And I became really fascinated by how they did this, how they built what ended up being, you know, within a decade, the world’s largest high-speed rail system, how they acquire the technology and how they built on top of that to create this sort of like truly like made in China kind of transportation system.And then how they did that repeatedly, not just for high-speed rail, but like for regular highways, expressways, for, you know, any kind of infrastructure.Okay, so that was sort of my foray into infrastructure and that was actually the focus of my dissertation. So I did field work actually for a number of years in China and also in India. I was based in Beijing and Delhi.And really it was like an enormous privilege to be able to travel around often by train across those two countries trying to understand their systems. And the railways were really useful for like understanding not just the railways and transportation, but understanding deeper political economy questions, like the structure of the governments, how their bureaucracy works, what are the main issues with building a mega project of that scale.And yeah, and so that for a long time was my focus. And then ironically, for an American audience, it was tougher to convince people that railways was interesting. I think most people were like, well, China can build high-speed rail because it’s the top down society and they just decide where to build and they build. I was like, no, no, it’s much more complicated than that. But it was hard to get traction.But then I realized like some of the same tools and the same patterns, some of the same institutions in China were also involved in boosting and accelerating development in key industries. So I mentioned clean tech, electric vehicles and batteries and solar, but also traditional sectors.And so I was really fascinated by this pattern that was, again, it kind of goes back to the New York Times piece. It wasn’t just one industry. It wasn’t just one company. It wasn’t just one state-owned enterprise, but this whole, like across the board effort to, you know, accelerate development overall.So that to me was so interesting—this process of industrial upgrading, which many, many countries are interested in doing. And that actually got more traction in terms of like, you know, I joke that the U.S.—every country in a sense is a developing country, right? There are areas where we’re trying to improve and areas where we’re trying to catch up.And I think now, you know, the question is like, what role can the right policies play in helping, say, the United States in a similar process? So that’s a long way of saying that it was—it was a long journey. But yeah, it’s an exciting time to study these topics because so much is changing. Every day it feels like.Grace Shao (11:30)That’s really good context and I think, you know, for me how I found your work was exactly because you had such a high level kind of view, a bird’s eye view of everything that was happening and how you were piecing it together.So I believe I came across one of your pieces on High Capacity, which is your newsletter on Substack for audiences who don’t know. You had this Venn diagram where you’re like, this is what China’s good at here, here, here, here, here. And this is what’s happening right now. And this is how it like actually relates to the current EV build out, the renewables, the AI, the robotics. It’s a very big ecosystem.And in some ways you argue that, you know, all these different sectors operate in parallel, whether is, you know, a top down direction or a directive, or it was organic, you know, growth, but they did grow in parallel. So therefore they’re now able to kind of find synergy and leverage each other’s strength, right? With LiDAR sensors, drones, robotics, all coming together.Could you tell us a bit about what that diagram really means? I’ll try to pull it up as well in the video. I think just help us explain that in a high level.Kyle Chan (12:31)Yeah, yeah. So what I try to capture with that diagram was this idea that it wasn’t just one sector or wasn’t just one area, one technology that China had been able to grow and foster maybe through industrial policy or some kind of state support, but it was this combination of these interlocking technologies.Now there’s a new term that’s coming into vogue, like the electric tech stack, or the tech industrial stack, or the electric industrial stack. And it’s really interesting because I think that really captures sort of this new paradigm that we’re entering.Those technologies—you mentioned a number of them—electric vehicles, autonomous vehicles, which for various reasons we can get into are built on electric vehicles. There’s strong reasons why those technologies go together. Lithium batteries, which also feed into drones, autonomous delivery systems.We can think about just regular consumer electronics, smartphones, but also more sophisticated robotics. I mentioned industrial automation as well. There’s a lot of overlap there.And then the big circle overlapping all of these is AI—different models, software platforms that might intersect with say autonomous driving and, I don’t know, even sort of the humanoid robotics world.So I just find it so interesting that China was making progress in a number of these different sectors at the same time, and progress in one sector would support greater development in another sector. So EVs and EV batteries grew up together in China.So the development of, say, lithium iron phosphate batteries that were increasingly inexpensive, that were increasingly energy dense on a, say, kilowatt hour per kilogram basis, that were safer—those developments within the battery world made Chinese EVs more competitive and more attractive overall.And then developments in China’s EV sector fed back into the battery world and also fed into other related sectors.And then the other big thing is that I really wanted to highlight companies that lay at the intersection of these different areas. So I think probably right now, one of the hottest companies is Xiaomi, right? When I was younger, Xiaomi to me was inexpensive smartphones and relatively like affordable, like household products, like air purifiers and things like that.And I think they had built up a brand and a very strong sort of customer base around this general idea. And what was so fascinating was seeing Xiaomi jump into electric vehicles and having such success with the SU7 and now the YU7 SUV, which both of which were like sold out for a long time and are very much in demand.And they have like incredible features, they have incredible performance, and they have smart driving capabilities. And so it just sort of like showed like, wow, this company that was originally kind of like a smartphone company could make this shift over into the EV world.And I would argue that it wasn’t just Xiaomi and Ledron’s entrepreneurship. I mean, a lot of credit goes to them for sure, but it was also because of this broader foundation that existed in China that allowed for this common supply chain ecosystem that would feed into these different worlds that would allow a company like Xiaomi to make that pivot.And you see it again and again. I mean, now you see—and this is where I came up with this term like the sort of Swiss Army Knife companies—now you see these companies like XPeng get into like a whole range of industries, right? So not just EVs, but also humanoid robots, drones, flying cars even.Huawei is probably the ultimate example of the Swiss Army Knife company, originally starting in telecom equipment, but then branching out into everything from sort of every aspect of consumer electronics—smartphones, tablets—into now AI chips. They’re a major player in semiconductors. For a while undersea cables. I mean, the list kind of goes on and on—EVs as well and smart driving, autonomous driving.There was just this like burst of companies coming out of China that could do all these different things and branch out into new areas very, very rapidly. And it was like shocking to me how big some of these bets were, like Xiaomi making a multi-billion bet on a new already highly competitive industry, the EV industry.But again, I think it all comes back to, in part, this foundation that was there in China—this like what I call these overlapping tech industrial ecosystems.Grace Shao (17:17)I have so many thoughts I want to throw out. One is, I think to your point on Chinese tech companies going into EV, it’s really fascinating because to your point, there’s obviously this price war and this crazy competition in China.But what I’ve heard from a lot of people who actually do purchase the Xiaomi cars and the Huawei cars and the Xpengs—they say that the technology itself is incredible. You have all the lights, the voice control, you have amazing AI-empowered functions. But that said, they’re not actually as good of a drive, like they’re not as smooth.So like if you pick a traditional OEM like a Mercedes or BMW, their voice control usually apps the crap, to be honest. Like they go off—like you see the reviews online because we were looking at family cars and it was like the reviews were horrible. Like they just get triggered by really random sounds, they can’t pick up like accents, you know, they’re not really good with other languages.And on the Xiaomi/Huawei/Xpeng/Zeekr side, they’re really, really good at this. But they’re not as good for driving yet. So it’s interesting that, like you said, what they’re good at in terms of day as the Chinese companies are the kind of technology that is quite recent and quite modern, but they’ve not really actually honed in on the craftsmanship or the, I guess, capability of building a really, really smooth driving car as Germans have—as they’ve honed the skill for over the last like four or five decades, right?So it is interesting where they’re good at. And then in terms of EV, I had a question you mentioned earlier. Most autonomous vehicles are now—sorry—most EVs are now being tried for autonomous driving. Why is that? What’s the synergy there? Why can’t old OEMs actually have strong autonomous driving functionalities?Grace Shao (19:00)Kyle, one thing we were talking about is the synergy between EVs and autonomous driving. Why is it that it’s better to actually build in autonomous driving functionality within EVs compared to like maybe traditional OEMs?Kyle Chan (19:12)Yeah, so there it’s basically because you get a lot more control and precision. And you can have things like steer by wire where rather than sort of mechanical steering, you can have basically a signal be sent directly to the transmission or directly to the engine or directly to the brakes. So it’s sort of all.And then on top of that, it’s helpful to have large battery capacity to handle sort of all of the different computing demands that would—including all the sensors that might be feeding data into the whole system.So yeah, you basically—the two almost always go together: having autonomous robotaxis built on top of EVs.Grace Shao (19:51)That makes a lot of sense actually. Okay, I never thought of it that way. You described China as building systems of capabilities and you kind of touched on this earlier with your Venn diagram. You talk about how China is not really just picking one winner or one winning sector, right?So how does actually EVs, batteries, renewables end up supporting each other? And then how does that actually extend out and spill over into the AI era with robotics, physical AI, or even the consumer AI products we’re seeing out there today?Kyle Chan (20:18)Yeah. So at one level, there’s an underlying driver here that’s almost not even specific to China per se. It’s something more about these technologies themselves and this broader convergence across them.So, I mean, I pointed out Swiss Army knife tech companies in China, but to be fair, right in the U.S., you have companies like Tesla that are going into many different areas, or even Google with probably one of the world’s best autonomous driving companies.So I think there’s something deeper happening here where there is this convergence of what we might have thought of sort of like lower end, like smartphone technologies or consumer electronics. Again, the lowly battery, right? Something so simple. This innovation in lithium batteries, making them cheaper, more reliable, and being able to scale up production—like that alone unlocked so much in terms of basically any kind of electronic device.And so I think that’s why we are kind of seeing this emergence of this like whole new technology cluster.And then for China, I think there is an awareness of the sort of synergies across different industries. And you can even go back further to China’s earlier industrial policy efforts, right? Take Made in China 2025, which came out in 2015.And some of the target industries there were chosen not just because they might in and of themselves be useful or important, but also because they had broader spillover effects. You think about things like telecom equipment, IT infrastructure, anything related to energy, or anything related to communication in general.Also CNC machines, right? So these are sort of like—they may not be like general purpose technologies in the way that electricity itself or computers are, but they might be sort of like multi-purpose technologies with broad applications in a range of different areas.And so by making that bet on those types of technologies, you know, whether or not you succeed in becoming a global leader in that area, it will help feed into everything else that builds on that kind of tech stack.So that’s what I see happen again and again, where for China’s approach to technology, where it’s not just about a single bet on a single technology, but trying to find these parts of the value chain that have large spillover effects and trying to support those—even if they in and of themselves might not be totally economically viable or the best businesses to invest in from a pure return on investment perspective, but they have these broader economic and technological spillovers.Grace Shao (22:57)Yeah, and I think to your point on like a lot of the planning from top down, it’s not like they were just more strategic in like picking the right track.But when I spoke to David Fishman a couple of weeks ago, he was saying China’s strategic planning on building electricity capacity is actually not because they foresaw like AI, the AI boom and data centers. You know, electricity is just simply urbanization actually was driving increase of energy demand as well.They knew that in the future, a lot of things had to kind of move from traditional coal to renewable to kind of actually even have the capability to power what is needed of the future. So it was a grander vision versus just like, I know AI is gonna come in 10 years. I’m gonna build up renewable energy and the renewable energy is gonna help power data centers. So it’s not like there are profits or anything. So that’s interesting to hear.I think I wanna understand how has that shift in ambition with like—Kyle Chan (23:42)Totally.Grace Shao (23:49)really China’s desire to move away from low wage, low margin, that trap into like higher value services and really like how has that shaped and driven the AI innovation that we’re seeing right now coming out of China.Kyle Chan (24:01)Yeah. So I think what’s really interesting is, on the one hand, you have like a lot happening within the AI sector itself. You have obviously this like very vibrant ecosystem of startups, of big tech companies, even down to servers and data center construction firms, and even the major state-owned telecom operators involved in data center construction.You have all these applications and developers trying to build on top of this whole set of foundation models, for example.And that’s sort of just within the AI industry, right? And then on top of that, you have the fact that in all these other sectors, whether you’re talking about manufacturing, you’re talking about biotech, healthcare, there’s a lot of investment and progress in trying to move up the ladder in each one of those industries.And so then you see people finding ways—either from those industries or from the AI side—trying to find ways to incorporate, to integrate AI.And actually there’s something that I really love about your work where you’re highlighting those areas where it’s not just about like the latest benchmarks on the latest models and this sort of like endless race, but about like, how is AI like actually being deployed? How’s it being integrated into existing services and platforms in a way that would really boost, say, drug discovery or would really, I don’t know, improve tutoring and education services for students.So I think that’s what’s so interesting here. And yeah, some of it might be supported by policy efforts. I think of some things like compute vouchers, for example, where startups—AI startups—might have trouble getting access to compute.So we’re not talking about like the Alibabas and Tencents of the world. We’re talking about the little guys who may not be able to afford to build a giant data center dedicated just for them. And then local governments might offer compute vouchers—subsidized compute—basically access to public infrastructure, essentially, to help them sort of get off the ground and have that little bit to deploy on and develop with.And so that’s an example of an area where you do have some government intervention stepping in and not trying to do it in a heavy handed way, but just trying to offer kind of like a tailwind of support.And ultimately, and this is one thing that I think is sort of special about software and the AI industry in general is, I mean, a lot of this is sort of like, you know, this creative explosion of different ideas from the private sector—from all these entrepreneurs—like trying to look in areas that are related maybe to their own areas of expertise or just kind of like scouring: where can we plug in AI? Where can we make improvements, even very small ones into existing industries?Grace Shao (26:45)That’s really interesting on compute bit, where I just met someone at Google in Singapore a couple of weeks ago, and he was saying that in some ways, the big tech in the West are actually operating in that capacity. Instead of incubating them and just taking another part of equity, and instead of just giving them capital, they’re giving them compute, essentially vouchers for these startups.So I guess my question is, how do you see the relationship between startups and big tech in China versus startups and big tech in the West? And in what way, I guess, in what way do the state actually play a positive role or negative role in all of this in the whole ecosystem?Kyle Chan (27:17)Yeah, that’s a question. I mean, in some ways it is a similar story, right? You have China’s own hyperscalers providing AI cloud computing services to a whole range of different players, you know, be they AI startups or, you know, large corporations or other, you know, maybe hospitals or other state-owned enterprises, for example. So that part might not be so different.But I think what might be a little bit different is the sort of like extra on the margin support that the government in China—or especially local governments in particular—might offer.Yeah, compute vouchers is one. Also these industrial parks where—and this is going back to like an almost an older model applied to like the age of AI—where, you know, AI startups, they still need office space. They still need help setting up a business. They might need help figuring out how to network with new customers.And those are other areas too, where local governments in China might play a more active role in troubleshooting, trying to bring startups up to speed, trying to connect entrepreneurs.And that’s something that I think is quite different than in the US system, where yes, you might have the large hyperscalers like Google or Microsoft providing that underlying infrastructure, that service for access to compute, but you won’t really see that kind of intervention or stepping in from the local government side, at least not so proactively by any measure.Grace Shao (28:46) Yeah, I think definitely the West what you hear more about is like applying for fellowships or acceleration programs within whether it’s VC funds or like you said the big hyperscalers. Whereas even in Hong Kong here, like the Hong Kong government offers startups like staff support, back office admin sharing, teams to share, then like even offices in like Cyberport out like, you know, in Pok Fu Lam.So like definitely the state plays a more active role and I think it’s kind of sometimes misunderstood by the West what that role means. It’s really many times it’s just like an incubator, a parent to someone, or even a mentor.So we talked about you living in many, many different cities across the world and you study industrial policies across different developed economies. You mentioned that you lived in India, then you studied obviously China. So from that perspective, that international global perspective, what do you think China has actually done differently compared to maybe other developing nations that started in similar circumstances maybe say three, four decades ago?Kyle Chan (29:46)Yeah, that’s a great question. So the thing that China has done that really has stood out to me, that really makes it so different than I would argue most other developing countries, is a very deliberate effort to basically build up industrial capacity and to move up the value chain.It wasn’t like, you know, if we invest in education, if we invest in sort of these general factors that go into development, that over time, you know, you would eventually sort of get there. It was like: how can we bring in foreign companies to form joint ventures with our own domestic firms and share that kind of knowledge? How can we build up world-class research programs and build interesting scientific collaborations with the Europeans or the Japanese or the Americans even?It was like: how can we try to like find those bottlenecks in the process?And I think this kind of goes back to your point where it wasn’t like, this is the one direction we’re going to make this huge bet and that’s what turned out to be correct. It was more of these sort of— you know, to use like a sports analogy—like kind of like the ground game, right? It was like oftentimes kind of more tactical things trying to backfill areas.Like let’s say for batteries, right? You need to have access to lithium. And so building a global network of lithium processing facilities and supply chains was really crucial to feed into EV batteries and then the EV industry itself.And yeah, so I think that’s something that I see other countries do like a little bit, but really for China, it’s sort of at a very deep level: this effort to not just sort of hope that you got most of the pieces right and then let the story unfold, but to proactively find ways to support industrial development and technological development.Grace Shao (31:44)What was actually, more personal, what was the most memorable thing living in Beijing and then commuting in New Delhi and then moving around the world so much over the last few years?Kyle Chan (31:54)Yeah, I mean, there’s so many things. Yeah, I mean, I can tell you from a research standpoint, I interviewed government officials in both countries. And I can tell you that it’s much easier to get access to government officials, at least at the central government level, in India.And so some of my sharpest memories are of long conversations over chai with railway officials in India, where I almost was kind of like a therapist for them in some ways, because they would have these complaints about the bureaucracy, about, frankly, their colleagues, about many thoughts about their country, about China, that they were just like very generous in sharing with me.And it was like really fascinating to kind of like see the world from their perspective. I think in particular in India, there is a bureaucratic elite, there is a civil service elite who are highly educated. They often have to go through extremely competitive national exams to sort of like get to their positions.And I think in some ways, I found this group of people to be both very proud, but also oftentimes very frustrated by some of the bureaucratic hurdles that they faced.On the Chinese side, yeah, like when I did get access to government officials, sometimes even getting access, right, like it might take a while to get to like a genuine conversation where people can kind of open up more.And it was so interesting because like I have my own theories about why say China was able to build high-speed rail so quickly, but it was always interesting to hear people’s own perspectives. To see like, what did they think was like really, you know, the thing that moved the needle.And I heard like sort of everything from, you know, that China is just a much more sort of like coordinated and aligned system across the board to cultural explanations to, I mean, you name it. But it was always fascinating to hear like from people within the system.So like to the extent that I was able to get that, and then I got to visit—I got to actually visit like the construction sites for like ongoing railway projects. And that was really cool.So yeah, I mean, there, there’s so many. Like the two countries are so fascinating and you could spend a lifetime in either one and feel like it’s not enough, in terms of exploring and getting to see different parts.Also, I did get to travel a lot. And like India, I don’t know if many people know this, like India has like the whole Northeast region, which is sort of like very distinctive culturally in terms of geography. It feels very different from the rest of the country.And, you know, like you could explore that whole area, or you could travel to the South and there’s like a big sort of North-South cultural divide that anyone from the North or South will tell you about.So yeah, it’s just like—I mean, these are sort of like continental-size countries. And I think they have that kind of continental-size complexity to them.Grace Shao (34:51)Definitely, I would love to visit India one day.Do you think what we’ll see India kind of become the next China, if you must put it that way? Because for many years people said, oh, Vietnam’s going to be the next China, right? And obviously all the talent, all the capital, all the interest in that right now moving to India, as well as even a lot of the supply chain, right? For a lot of the big companies in Europe as well as the US, what should we expect of India in the next maybe decade or so?Kyle Chan (35:14)Yeah, that is like the big question. And I will highlight some factors that are very much in India’s favor, and then I can point out a few challenges.So in general, the India story is like really fascinating because if it weren’t for China’s high growth story—like if you kind of remove that from the equation—India would have been the envy of the world.I think in many ways they have already proven to be able to have high rates of sustained growth over multiple decades now. So like in that sense, they’re already getting there in some ways.The other things that are really working in their favor are: there is a big focus on manufacturing, on trying to build up industrial capacity in a way that is very reminiscent of China. There’s a huge push in industrial policy targeting areas like consumer electronics, the automotive slash EV industry. There are big ambitions for the semiconductor industry in India.And the third thing I would point out is that there, up until recently, was a bit of a geopolitical moment for India with different companies trying to move away or diversify from China to some extent, diversify their supply chains.And this was really quite an opening for India where you saw, like for example, Apple was able to shift a fairly large share—I think up to like a quarter—of its iPhone assembly was shifted to India. And with that also began a process where some of the suppliers would start to follow. And so I think that was actually honestly, to me, surprising how quickly that happened.So those are some factors in their favor. But on the other hand, there are some very deep structural constraints. And one is that the bureaucracy, especially around anything related to labor, is very, very tricky to navigate.And I think for companies who want stability, who want sort of like a certain stable business environment, it’s harder to operate, and there’s a lot of regional variations. So some of the southern states like Tamil Nadu, Karnataka tend to be seen as more business friendly, open to foreign investment, open investment of many different kinds. And that is where you see the electronics industry really growing quickly.But overall, there are some bureaucratic issues. And then at a more fundamental level, something that China has been able to do is often reform its own internal organizational structure—sometimes pretty quickly, sometimes shockingly fast.And sometimes it takes a while, like the railways did take a while, the railway ministry. But for India, I think there are some problems that everyone knows exist in a policy sense, but nobody knows how to address and break, say, like a political deadlock. That happens a lot. It probably also happens—I mean, I know it does happen in China as well. But I do see it more clearly and especially in the areas that India is trying to target for growth.Grace Shao (38:10)And do you see actually a lot of the Chinese companies we talked about earlier, like the Xiaomis, BYDs and Huaweis of the world—are they exporting to India or even exporting their know-how and their supply chain and manufacturing? Because I know a lot of these Chinese companies have been doing that, like moving their talent, their know-how and also selling to consumers in Southeast Asia. But what is kind of the South Asia market looking like for these Chinese companies?Kyle Chan (38:35)Yeah, I think from the Chinese companies perspective, they would be eager to reach one of the largest markets in the world, right? And then South Asia more generally.And I think you see other areas like Pakistan and Nepal—Chinese, say, EVs or other smartphones or other consumer goods—like really taking off.India and China specifically, though, have a bit of rocky geopolitical relationship. And so that definitely cast a chill over cross-border investment and business flows for a while.And more recently, it seems like things are warming up, maybe gradually, and I think this is taking years. But I think there’s a recognition, certainly from the Indian side, that it helps to bring in that expertise and know-how—to have Chinese engineers, technicians, and managers participate in India’s own industrial development process.I think, I don’t know if we’ll see those days again when you had, say, Alibaba invest like, you know, hundreds of millions in PayTM, which was India’s sort of equivalent of Alipay. I don’t know if we’ll see those days again, but maybe there is a sort of warming up of relations between the two countries and we might see greater sort of business flows following that.Grace Shao (39:48)Yeah, because it seems like India has their own ecosystem of like FinTech as well as their own ecosystem of social media. So it’s not like they really need China’s software export.But I think on the hardware side, it was really interesting reading Patrick McGee’s book and he was saying how back then it was the Taiwanese and the Americans that actually had to—well, the Americans first trained the Taiwanese and the Taiwanese came to train the Chinese.And now it seems like if the supply chain is moving over to India, I’m sure the business will be business and, you know, it would make sense for the business to actually have the Chinese now train the Indian factory workers to really take over that new labor demand, right? Yeah, but geopolitics aside, that is.Let’s talk about AI quickly. You know, in your conversation with Heath Yap—also I love his show and a good friend of mine—really grateful for Keith connecting us actually.In your conversation with Keith, you said that you’re always careful to hedge. I kind of chuckled when I heard that. You’re talking about China’s rise and China’s rise in AI. You’ve obviously been quite vocal talking about how China has done some things right. But yet, obviously, as an American, you are saying it from a perspective of what we can learn as Americans.And what are the scenarios where your own projections—China’s technological rise—don’t pan out or that could derail this story, right?Kyle Chan (41:00)Yeah, that’s a really good question. So there are some ways that this story can turn out differently.I do think that China has been able to benefit for a long time from especially international partnerships. And I think the extent to which there might be skepticism or even suspicion about those sort of international partnerships—that could affect, say, research collaborations or business partnerships across border.At the same time, though, I guess at this point, I can point to areas outside of AI where there is a lot of interest, say, from German or European automakers in partnering with Chinese ones, trying to learn some of that know-how and technology. But that’s a big wild card.And then another big one is also the extent to which there is a bit of a trade backlash, right, to Chinese exports and to what extent other countries might push back and maybe put up more tariffs or maybe even just demand greater investment and localize manufacturing in their own countries. So that could change things, not necessarily in a worse way, perhaps, but it could change things in some ways.And then I do think that the relationship with the US is really crucial. And I think for China’s development, the instability in the relationship—I mean, it doesn’t help anyone, I think. So that’s sort of another wild card in all this.And then from the US side, I mean, it’s sort of interesting to think about American AI development right now, because I literally like every day my feeds are like talking about bubbles and talking about like, you know, circular investment deals and things like that.And at the same time, like there’s still so much enthusiasm and the data center build out is like seems to be still going full blast. So here’s like where I will offer like a hedge perspective where it could go in many different directions.It could be a bubble and we might see a pullback and it might turn out that these investment deals just don’t pencil out. The economics just don’t work out.But it could also be that we’re not—that this talk about AGI, I don’t know if like AGI per se will be achieved—but that broad-based artificial intelligence like computers that can do a huge range of tasks, not just coding, which is already impressive, and not just writing and things like that, but really sort of a broad range of tasks—that could take off.And the United States huge investment in compute capacity could give it an edge in the long run.So those are different ways things could pan out. And then the other wild card is like embodied AI and robotics and how that will pan out.Because I think there’s like different approaches being taken by different companies, by different countries. I think some companies are looking to a version of AGI for robotics, kind of a universal foundation model for robotics that can operate across any different sort of hardware platform.And other companies—I think about like Unitree—are focused on just deploying fast, like building tons and tons of like quadruped and humanoid robots now that have become kind of like the new hardware platform for many robotics developers, including in the United States.So there’s many different ways things can go.And then like more recently, there’s like, I have this tweet about like highlighting like street cleaning robots and just like very practical, like immediate real world use, right? So that could be another area where, you know, maybe we won’t get the AGI robots, but maybe we will have many, many different types of robots doing more specialized tasks.And that could be transformative for either country. These are all areas that I’ve followed closely. Yeah, at this point, if I had a crystal ball, maybe I would be in a different job.Grace Shao (44:57)I think he definitely touched on something I feel like it resonates with me. Like is when I speak to people in the US—not really just the West, but really just in Silicon Valley—I feel like exactly to your point, there’s such a polarized kind of take on where AI is going.It’s either it’s such a bubble or it’s going to burst. It means nothing. This is complete fad, right? Or it’s—or even, you know, people are like, this is going to ruin our lives. We must stop it right now.Or it’s a very polarized view, which is like, OK, this is like the best thing that’s ever happened to us, like, you know, AGI or nothing.And in China, when people ask me about what’s happening on the ground, it does feel like—I’m not sure pragmatic is actually the right word anymore. I don’t like using that word anymore because I feel like it’s now overdone. But I do feel like it’s a bit more, I would say centered.As in everyone kind of feels like it’s just another wave of technological advancement. Whether how it might pan out—to your point—it might not be AGI, it might not be this crazy intelligent being that will take over our thinking capabilities, but it might be something that will transform how we humans interact with each other, how we work, how we actually increase productivity.And I think it might have some—it might lean on the fact that frankly China really did see this kind of technological advancement driving productivity gains only just recently, within the last generation. Whereas in the US, you haven’t really seen this kind of crazy gain for the mass in a while, right? Since the Industrial Revolution. So people don’t feel it as much and there’s more fear around it.Now I have another question on this. Just to kind of wrap everything up: you look at a lot of the industrial policies and you look at how the state works with the private side in the AI sector right now in China. How much are we seeing that is actual genuine market pull versus state-driven push?I think you kind of alluded to the fact that like Chinese government does work with helping talent and raising capital. But right now with the big names—like the Unitrees, the UP techs, the deep CXC, Galabots and the whatnots, right?—that we’re seeing, are the state behind this or like the West often seems to think that way. How do we understand that?Kyle Chan (46:56)Yeah, so I think it’s a complex relationship. So I don’t think it’s sort of like a top-down, the government is sort of like dictating what direction these companies need to take.And I think there are probably national priorities where I think Beijing or local governments would want to see companies focus on more.And ironically, like there’s a very strong convergence right now on wanting to make progress on AI more generally from the public and the private sector.And then I do think that some of it comes down to problem solving on the ground. So this is something we talked about earlier: trying to figure out what issues the private sector is running into, what issues certain firms are running into, and then trying to troubleshoot and support them in some of those areas.And then, yeah, overall, sort of like offering this broader roadmap or this broader set of like this broader package of sort of like a policy support, whether it’s sort of like a robotics-specific national strategy plan.And part of it is just about even signaling that there is state support for these sectors, that these are areas where if you venture in as a private business, like you will not have all your problems solved, but you could at least have ways to have some of your sort of like more day-to-day issues addressed.Grace Shao (48:17)Kyle, if you look at the AI sector today in China, how much do you think we’re seeing genuine market pull versus state-driven push? And especially now we’re also seeing the AI plus policy being pushed out. Is that diffusion driven by the state or is it driven by the actual entrepreneurs, especially amongst the Unitrees, the UP techs, the DeepSeeks and the MiniMax, Moonshots of the world?Kyle Chan (48:39)Yeah, so that’s a great question. And I think what’s interesting about this moment is there’s a strong alignment between the public sector and the private sector, where clearly Beijing at a national level, and then many of the local governments want to see a booming AI industry in China.They want to see China as a whole make progress in applying AI to all different parts of life—support economic growth and improve social services and improve education and all these good things.And at the same time, you have so many different private sector firms from the big tech companies to the smaller startups really jumping in and also eager to innovate.And I think one thing that’s really sort of exciting about all this is you can really like drill down into like subsectors or different aspects of China’s AI industry and just see how vibrant the startup ecosystem is now in terms of all the different, say, new coding tools that are emerging or different efforts to try to expand overseas or interesting ways to integrate AI into e-commerce or social media or EVs or advanced manufacturing—you name it.And so right now, I just think that there’s just like this huge sort of creative explosion of different ideas and an eagerness, especially in this sort of post DeepSeek moment, to try new things and take that risk.Grace Shao (50:04)Super interesting. Kyle, I really appreciate your time today. And for listeners, we actually had a few technical glitches today. That’s why we’re going to have to chop up the conversation a little bit. It might sound a little jumpy than my usual podcast.My last question for you, which is a question I ask everyone: what is one differentiated view or non-consensus view you hold? I guess your New York Times piece was a pretty strong opinion and you really publicly put it out there. But is there anything else you think that, you know, you think differently or you have a different kind of view on that compared to your peers maybe?Kyle Chan (50:35)That’s a question. I do think in general that my take—maybe it’s becoming more mainstream—that there is a role for very creative and very thoughtful policy intervention in supporting strategic sectors, in boosting technological innovation.I think that’s something maybe perhaps widely accepted in other countries, but at least in United States, still something that a lot of Americans are not quite comfortable with.And there are certainly risks along the way, but I think also to keep in mind, there are risks of not doing anything and of not trying to support R&D, scientific development and all these different things and to leave it purely to the market. That’s sort of my slightly left field take.Grace Shao (51:21)Thank you so much, really appreciate your time and your insights again. Thank you, Kyle.Kyle Chan (51:24) My pleasure. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  AI Governance From Brussels to Beijing: George Chen on APAC’s Different Path 16.12.2025 52λ

  Most AI policy conversations still orbit around Washington and Brussels, but Asia-Pacific is already writing a very different rulebook. In this episode, I talk with George Chen, Digital Partner at The Asia Group and former Meta policy executive, about how AI is actually being governed, built, and deployed across APAC, China, and the global south.George traces his own path from journalism to big tech to advisory work, and uses that vantage point to explain why APAC is not “one market”—and why the EU analogy breaks down almost immediately. Countries like Japan, Korea, Singapore, and China are leaning into AI as a tool for economic recovery and industrial upgrading, often taking a much more pro-innovation, pro-growth stance than the EU’s more precautionary approach. At the same time, Southeast Asia is becoming the physical backbone of the AI build-out: Singapore as HQ and regulatory hub, with Malaysia, Indonesia, Thailand, and the Philippines hosting the data centers, power, and connectivity—along with all the local tensions that come with that.We also get into what “responsible AI” actually looks like inside a company. Beyond the buzzwords, George breaks it down to three pillars—security, safety, and privacy—and talks through how mature players like Microsoft or Meta build these into product design from day one, versus the reality for startups trying to ship fast with one lawyer and a single policy person supporting multiple markets. He also makes the case that fragmented regulation and the lack of international standards are becoming a real tax on innovation, especially outside the US and EU.Another big thread is the emerging US–China competition over AI governance itself. It’s no longer just about who has the best models or chips; it’s also about who exports their rules, norms, and defaults to the rest of the world. The US is pushing an “America-first” innovation and safety model to allies, while China is pitching AI as a kind of public good to the global south—combined with a more cost-efficient, top-down deployment model and very strict cyber and real-name rules at home. George argues this divergence is already shaping how content, deepfakes, and AI-generated media are treated in different jurisdictions.We talk about the local edge of Chinese models—why in places like Beijing, models such as DeepSeek can be more useful than ChatGPT or Gemini for everyday queries because they’re trained on more localized, timely data. From there, we zoom out into the new AI talent map: countries like Indonesia, Vietnam, Kazakhstan, and Uzbekistan trying to position themselves as low-cost AI talent hubs and “back offices” for global AI companies as coding gives way to prompting and applied ML.We close on a more philosophical note: should AI be built as a subordinate assistant or a true partner? George shares his uncertainty here, and we talk about what happens when we give AI more agency, emotional intelligence, and continuous workloads. At some point, the conversation shifts from safety checklists to ethics, culture, and even “digital colonialism”: whose values, whose norms, and whose worldview are encoded into the systems that end up mediating how we see the world.In today’s world, there’s no shortage of information. Knowledge is abundant, perspectives are everywhere. But true insight doesn’t come from access alone—it comes from differentiated understanding. It’s the ability to piece together scattered signals, cut through the noise and clutter, and form a clear, original perspective on a situation, a trend, a business, or a person. That’s what makes understanding powerful.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.For more information on the podcast series, see here.AI-generated transcript.Grace Shao (00:00)Hey George, thank you so much for joining us today. I’ve been really excited and waiting for this chat. You know, you are a very busy man. You’re constantly traveling. I can barely reach you in Hong Kong. So really appreciate your time today. Sit down with me and share your insights with my followers and some of our listeners. To start with, you’ve worn many, many hats. A journalist, tech executive, policy advisor, and now a partner at the Asia Group where you advise a lot of force, you’re probably helping companies on, I believe, geopolitical positioning, right?George Chen (00:29)Thank you. First of all, thanks for the invite. It’s quite an honor to join a growing cohort of guests for your program. Really happy to have a discussion about tech and policy issues because I think you’re right. My first 10 years in media, similar to your background, and most recent decade, I work very much on the intersection between technology and policy.My biggest takeaway from my last job at Meta, one of the platform operators in the world, is sometimes we very much focus on technology development, like the breakthrough, while the resources for policy support are actually quite limited, especially in the Asia-Pacific region compared with the US. think for all the...Big tech in the US, given the politics domestically, they have to do a lot on political and policy part. But for Asia Pacific, the policy work, compared with other investments, like in data center, technology, hiring of engineers, it’s still very, very, very understaffed, under-resourced, and sometimes under-appreciated. This is why we need to...address some concerns about policy issues as we advance the technological part. Because I always tell my students, tell my friends, tell my partners that the key challenge, even you have CharGBT 5.0 or 6.0, the key challenge is how to get the government to understand new technologies and also get the users to have more trust in those new technologies. Otherwise, nobody use it, nobody trust those things. And that makes them.Grace Shao (02:15)I think that’s super helpful. A lot of times when we think about policy or safety issues, we think about it as like a siloed part of the ecosystem. But really like exactly to your point, like, you know, we need the developers to understand the concerns of the users. We need the users to understand the safety risks of the products. We need the regulators to understand what it means to implement these like technology throughout our economy, right? So there’s it’s like, it’s actually all interrelated.I think today to start off with, let’s like go into big tech, just give in your background with Metta, working with a lot of these big tech companies. You’re based in Hong Kong for the listeners, but actually work predominantly for American big tech companies. What is like the, I guess, the fundamental feel right now as we see the evolution be from a social media company for AI to AI of focused company as this is now the forefront of their strategy.George Chen (03:11)Right, so for the Asia-Pacific region, it’s big. I always try to explain to my clients and friends, when people talk about Asia-Pacific, the first gross perception, perhaps from Western perspective, is, okay, treat Asia-Pacific like the EU, right? But EU is a single market. They have very much shared the language, English, also one currency and they have the European Parliament to pass legislation for EU member countries. Asia-Pacific is far diverse, far different, and much bigger. So it’s hard to just copy whatever works in EU and then let’s also do it in APAC. Using AI regulations as a clear and classic example, you know, you is the first You know government, you know to have the world’s first AI act, right? But the so-called the Brussels effect didn’t really happen this time in Asia Pacific countries You didn’t see like all the countries, you know, like Singapore or you know Japan to quickly follow up on You know to have a similar like a risk-based approach or penalty focused approach to AI, right? Instead, you know if you look at Japan. They are very much welcoming. Japan declared to be, they want to be the most friendly open country for AI developments. The first data exception for AI testing was actually in Japan. And then Singapore followed, and Hong Kong’s also not considering, right? So APAC took a very different regulatory approach to AI versus EU. I think this is something all the American tech companies have to realize. It’s not like America leads technology and then EU matters because of the special relationship between US and EU. So as I mentioned at the beginning, the resources for public policy work are very limited in AIPAC, but EU still enjoy a lot of resources, this English-speaking market that has lot of political connections. And then Asia-Pacific, when it comes to policy enforcement, like policy support it feels more like a third country, overall speaking Asia-Pacific as a whole. So there’s still a lot of educational process, the learning curve for big tech, largely from the US to understand what are the challenges, what are the opportunities in the Asia-Pacific market. However, I also need to highlight for many big platforms, Asia-Pacific is actually not just the largest market by internet users for American tech companies, for almost for all of them, right? You know, in terms of user base. It is also a very important revenue source, know, the source of revenue for those American companies. So now you see the imbalance, right? You you make a lot of money from Asia-Pacific, but the support you give to Asia-Pacific is quite limited, know, compared to in the US ⁓ and EU. So the learning curve is there.American tech companies want to have a more sustainable development and want to have a more constructive relationship, sort of a more constructive partnership with Asian governments. I think there’s still a lot of work to do.Grace Shao (06:31)I think that’s really helpful to help listeners understand because sometimes people also approach me, they’re like, what’s APAC? I’m like, APAC is gazillion different markets and it’s actually so fragmented, right? And I think people sometimes misunderstand it kind of similar to like what you said. They think it’s like a EU. It’s not like actually there’s no consistency in currency. There’s no consistency language or no consistency actually even income or anything. So it’s quite scattered. that sense, I actually want to ask you, you mentioned something just now. George Chen (06:39)That’s right.Grace Shao (06:58)Japan and Korea this time is taking a more proactive actually approach as the countries themselves are taking more proactive approaches to really embracing AI and you know actually compared to EU’s more wait and see or more protective measures right which is not very yeah not not not what they usually would do what do think the trade-offs are actually in that sense do you actually think that means we are seeing more innovation or more technological breakthroughs or even economic diffusion of the technology right now in Japan and Korea.George Chen (07:30)Yeah, yeah, let me put it this way. So AI technology, you know, we believe, you know, still in the very early stage, right? Even you talking about, you know, trying to redefine Polisero, you know, but, you know, if you put that in the overall development for AGI, you know, we are still very much under the, in the early stage of the curve. So for Asia Pacific region, yes, it’s diverse, you know, but we can still see some sort of patterns, similarities in terms of different AI strategies. At the Asia group, my firm, we did a research paper on the different regulatory approaches to AI governance in the vast Asia-Pacific region, from Australia to even in Mongolian. Long story short, you are right. Some countries in Asia-Pacific take ⁓ a more economic benefit focused approach, right? Take a more innovation focused approach. Countries like Japan, Korea, Singapore, they want to see how AI can help them to drive economic impact, right? It doesn’t mean like they don’t care about the safety, the security issues, but they want to have certain flexibility, to encourage more startups to succeed, right?in to a certain degree, actually maybe too many surprising because China is very well known as one of the strictest internet market in the world. Basically, none of the American, very few, I will say, like very few American tech companies can really succeed in China. The only two exception in my mind are like Tesla and Apple. But they are more like consumer related if you touch on content.We talk about Google and Meta, that’s a completely different story. But even so, China at this time is also taking a more pro-innovation, pro-economy approach to AI development because this is a very top-down approach because President Xi saw the success of DeepSeek and he basically wanted more success stories like DeepSeek. Japan and Korea are in more or less the same category, like pro-innovation, pro-economic recovery. For Japan,I talked with my friends and colleagues in Japan. The sentiment in Japan is like, we’ve lost 30 years, guess, three decades in terms of economic recovery. This is like our last chart. And Japan has been quite strong in robotics, those fundamental technology development. So that’s the sentiment in Japan. We have to grab the AI opportunity. In EU, have to say, part of the reason why EU is so keen to develop regulations, legislation in recent like five to 10 years. In my view, some may argue and disagree. I think the EU does come with a sense of protectionism, right? Because if you look at all the market leaders, you name it, OpenAI, Google, Microsoft, AWS, all of them are big tech from America, right?I remember there was a chart to list the top 10 most advanced AI models. There’s only one model from EU, actually from France. The rest are from the US and China. So that tells a lot. If you are the EU regulators, look at from a competition perspective, you will more or less have a sense of anxiety. And then you will look at all those big tags like, no, we need to do something, like a country that pays in the name of safety and security. I’m not blaming EU regulators for doing it. But in the meantime, we also hear more and more concerns, even from the state heads, like French President Macron. He’s concerned that tough regulation in EU on AI will harm innovation in the EU rather than help European startups.Grace Shao (11:14)I think we can double click on China later. It’s going to have its own special segment for sure. China is just such a big story. But for some context for lot of listeners, Meta and Google, the likes of these companies actually do exist in mainland, but they mostly only have their ad services there. So basically they help enterprises with their ad sales to the West. But to Georgia’s point, they’re not really operating at the full capacity that you would see them elsewhere in the world.George Chen (11:33)That’s right.Grace Shao (11:39)Now I do want to kind of finish up on the APAC kind of narrative and then the APAC focus right now, which is for ASEAN right now. Let’s set apart like South Korea and Japan and China, just the Northeast Asian countries are frankly economically much more, you know, like developed as well as more economically focused, right? For ASEAN right now, especially since I just went to Singapore last week, it’s really interesting. Like we basically have the players, like you said, OpenAI, Google, Meta, all of these. Well, APAC headquarters based in Singapore, even the 10 cents and the bite dance of the world, right? However, Singapore is tiny, like just in terms of size and its resources. So what we’re seeing is they’re extracting essentially all the compute energy data centers, connectivity, any of the infrastructure you need to think of actually to Malaysia, in Malaysia, in Indonesia, in Thailand, even they’re building them out over there. How do we actually understand this right now? Is this a net benefit for these economies? Or is it actually really hurting the local economies and, you know, in some ways exploiting them and really just only serving the companies based out in Singapore? How do we understand that?George Chen (12:45)That’s right. So let’s talk about Southeast Asia. It’s complicated. When we’re talking about APEC, actually the most complicated part, I think it’s like Southeast Asia. Because when we talk about Korea, Japan, China, even China is a socialist country, but in terms of economic models, there’s a lot of elements related to capitalism. So those are the most economic economics in the Northeast Asia. Southeast Asia is very diverse, very different from each other.Singapore is like the exception, the most advanced economy in South East Asia. But they come in terms of population, the user base is pretty small, like 4 million, 5 million population, even smaller than Hong Kong. You’re right, a lot of the tech companies, even before AI become a trend, they talk about like Meta, Google, Apple, they all had their headquarters in Singapore. It has really become the hub for big tech over the past 10 decades. Unfortunately, Hong Kong, thank God that we still have big banks like JP Morgan, Goldman Sachs in Hong Kong, we remain as a financial center. But in the aspect of tech innovation, you have to give some respect to Singapore. They did very well to attract those tech headquarters. So this also became, you are right, sort of a point ofI don’t know how to describe it. Some of the neighboring countries are jealous, certainly jealous of the success in Singapore, right? And then countries like Indonesia or Malaysia also wondering like how to get the benefits from the fact that all the big tech have their headquarters, regional headquarters in Singapore, right? But if they only care about the relationship with Singapore or in government, because they have headquarters in Singapore and their neighboring countries will not get any benefits, Malaysia actually founded their own ways in the regional AI race. And their offer is data center because of the stable supplies of electricity, relatively much cheaper labor costs and land costs and overall cost for data center operations. So this is why Malaysia got a lot of attention from Big Tech too, like AWS, Microsoft, they all made huge investments in Malaysia. Not AI, R &D, maybe yet, but first our data center. In the AI industry, we have a popular saying that AI is like electricity. Sam Altman said that. Basically, this is like the new kind of utilities for everyone’s life, right? But to develop AI you also need electricity. You need a lot of investments in infrastructure. This is why Malaysian already stand out and Philippines too in a way, as sort of the cheap, reliable alternative to data center investments in addition to Singapore. Everybody complains about Singapore in terms of living costs, even like how difficult it is to get work committed in Singapore these days. Even you have a call like qualify the job, but it doesn’t mean like you will get work permit immediately. Actually, in comparison, Hong Kong is doing quite well to attract the talents more easily these days in the tech and financial space. Back to the AI governance issue, yes, Southeast Asia also took a very different approach in comparison with Korea and Japan. think Singapore is an exception. Otherwise, if you look at the countries like Indonesia, if you look at countries like Vietnam, they still take a very more security-focused ⁓ approach, especially Vietnam, given their political system, right? So, Meta used to, and I think Meta still have a lot of problems in Vietnam. One of the key issues is about content and moderation, right? There’s a lot of human rights and similar struggles, Thailand too. So those countries, I feel like the sort of the older problems from the social media era was not really solved yet.And those problems will be brought into the AGI era. And when the government look at the AI, their first question is, okay, so how can I prevent people from using AI to cause any unnecessary trouble, which means like a social instability, right? So that will be the same older problems facing big tech companies. And that tells you a lot when those countries look at AI, they still come from very much a security focused on mindset.Grace Shao (17:09)That’s really fascinating because actually when we were just talking about the infrastructure build out on my end, really just I’ve done some research and writing on, you know, the Johor build out and the over capacity with data centers right now. And the unfortunate cause that’s just like the local infrastructure is not able to actually support the the rampant build out. It’s actually affecting the livelihood of people. Right. But your point is really interesting. I didn’t really think about it that way. It’s actually for the from the perspective of these big tech actually.It’s to prevent bad actors using their technology to actually propel even further, like, you know, bad, intentional, harmful content, right? And then essentially, like you said, cause social unrest that would really be very troublesome for the local government. So I guess from the policy perspective, from the social media era. But what would be something different? What would be something that, you know, big tech will have to start thinking about that they didn’t even have to worry about before.George Chen (18:03)Right. Well, you know, as I said, know, lot of the older problems from social media era will remain in the AI era, such as misinformation, know, skin, political speech, you know, especially for countries like Vietnam and Thailand, the real content. It’s always, you know, when I worked at the Meta, you know, those South Eastern countries are always considered as like a high risk countries, you know, when it comes to content policy risks, right?On the other side, Vietnam, Thailand, Indonesia are much bigger. They are also smart. They also consider AI as opportunity. So they are thinking, how can I use AI to train the next generation of talents, digital talents? Those countries also have relatively younger demographics. So there’s a lot of smart kids who can get on AI and then to learn. So I think that also posed the opportunity for partnership for those American tech companies. Can we do some training program for the purpose to grow the next generation of AI talents in those countries? I think those governments will be very much welcome those initiatives. And this is not just happening in Southeast Asia. You may know somehow I also have my exchange of career experience in Central Asia. I can tell you even countries like Kazakhstan and Uzbekistan trying to focus on talented developments. Because they believe, if you think about learning how to code 10 years ago, this is actually quite an expensive experiment. You need to get professional tutors. You need to get long hours to learn one language. When I grew up, I learned like, I don’t know if you know, we started with Microsoft, the DOS system, and then C12, no one talking about it.So it took like a year to just get a basic sense of those languages, right? But like AI, you don’t need to learn ⁓ the code. It’s more important for you to understand how to write a proper talk. So those countries like Uzbekistan and Kazakhstan also catch up trying to be the back office for big tech to train, to grow the basic, like the junior engineers. So hopefully they can get some basic work done in those countries for cheap labor cost reasons rather than you need to hire all those engineers in Silicon Valley. And I think that posed the same sort of opportunities for Indonesia, Vietnam, Thailand and other Asian countries.Grace Shao (20:30)That’s really interesting. So there’s like a reshuffling of talent and then also like just the talent strategy is actually changing from the social media era or just like the big tech era. I want to kind of look at responsible AI. So we hear the phrase a lot, right? Responsible AI, AI safety from your experience right now.What does responsible AI actually look like inside of a company and what changes in org charts, KPIs, or decision making when we’re talking about responsible AI? What are the metrics we must track?George Chen (20:59)That’s right. Okay, so you mentioned that I wear a lot of hats. You I don’t want to speak like a professor, but I do teach a course at the University of Hong Kong and the Tsinghua University. My course is about digital society and governance. One of the lectures is actually about AI governance for corporates. So responsible AI is a term, you know, very popular, not just in the tech industry, but you now hear more and more just in business in general, right?It’s, in my view, responsible AI is something like the privacy statement, right? You know, for different companies, you know, when you go to a website now, like the privacy statement already become like the very normal thing, right? You know, when you use a service, you know, have to get, you know, they have to get the user consent first, and they need to tell you, you know, what kind of data they’re collecting for what purpose. That’s the privacy statement. Every website, you will find, you know, a privacy statement. Responsible AI is similar.So the government is doing their job ⁓ from regulatory perspective, from self regulatory perspective, the government work with NGOs and associations to have an industry code. But for corporates, responsible AI is kind of like the business led principles. I want to use Microsoft as a perfect example. I think Microsoft is leading the way how business can take a more responsible, sustainable approach to AI. Microsoft is responsible for AI. They call it the trustworthy AI, but it’s just the name change, more or less the same. Microsoft very much focused on three pillars, and I believe many other AI type companies focus more or less the same. First is security. You have to have a very secure AI system. That’s the basis. That’s also where the user tries to come from. Second is about safety you talk about online safety, particularly for those more vulnerable groups like children and women, how to address those issues. Again, the same social media problems like harassment, online safety, even suicide prevention, exists, if not get worse. The last one, at least, is privacy. That’s easy to understand. So, safety and security privacy. The three pillars are the key foundations for responsible or trustworthiness or other names. When we talk about the process for big tech or just traditional business like Starbucks, when they want to implement AI in their business, we have a massive means called the privacy by design in the social media era, which means privacy should be the first thing to consider when you develop a product. This is like a rule. ABC like a 101 for any product manager, right? You know, when I worked at the Meta, we always got a reminder like, you know, the engineers, right? It’s not like you have a great product idea and you talk to everyone and finally you think about, okay, I should talk to my privacy lawyer. All right, you should do it the other way around. The first of all, you should talk to is the privacy legal, the privacy team, right? Responsible AI poses a very similar approach. The first thing, when you develop an upgrade or a new service backed by AI, you should think about whether you can tick the three boxes, security, safety, and privacy for the AI services and product you’re going to launch. Microsoft has set a very good example when they’re developing the co-pilot. That’s their AI platform. for you users. So I hope that can give you a very rough sense of what responsible AI is about.Grace Shao (24:39)I think what you mentioned just now that stood out to me is that a lot of these big tech companies like Microsoft or Sell, they have very mature, legal, and safety teams in place, right? So it’s much easier for the developers to actually tap into their know-how and their knowledge. And obviously, like you said, an extension of how they use their regular content as well as not just content moderation, but also just product safety. But for startups, I don’t know if you work with them at all or not, but like,I just the proliferation of AI tools right now, right? It’s like, it’s very, it’s very crazy right now. Basically, like you also kind of hinted at this where like, you know, developing a new product is so much easier than it was say 30 years ago. It’s not only that, like, you know, the language of coding has made it easier, but now we have a jented coding tools, right? So you can have vibe coding, whatnot. How do we actually understand product safety and like responsible AI when we start talking about new products within these startups. And also my question is on a broader like picture, how do we understand responsible AI in a big market like China where a lot of products are consumer facing AI versus maybe the US where it’s a lot more enterprise facing. Can you kind of give us some color on that?George Chen (25:55)Right. So first of all, startup, yes, you we do have some startup clients. I’m very glad that the startup clients we work with in the tech sector are very much, you know, either backed by some leading figures in the Central Valley or by global VCs. So I think that they do have, you know, like a stronger internal compliance control. Right. And over the years, I think all the big tech, you from know, met up to Microsoft, you know, to other companies. I think all the classic incidents, know, the lessons, remember, you know, when Mark, when Mark Zuckerberg had to apologize, you know, you know, the Cambridge Analytical incident, right? It feels like a not too far away, you for people who had short memory. I think that those incidents that did, you know, ⁓ serve as very good lessons, you know, for those, you know, I will say like a more US-funded back than startups. I think their goal is clear. If you want to get listed on Nasdaq someday, you’d better do things very right from the beginning. There are some naughty boys, cases from China. You probably noticed there are some AI-cub startups from China.Like grab the content from Disney, know, Parliament, know, Sony, right, you know, to make those funny, like the AI, you know, effects. But in fact, it was like a serious violation of, you know, IP prototype content, you know, but those start up like, I don’t care. I just like to have fun. Like, let’s see how it goes. then suddenly, you know, they got like a 1 million to 2 million, you know, and then to 10 million users within a week. So, but they’re not going to go far away, right? There’ll be like a long series of this and that. So this is not the right approach. I do think that startups need to be very clear about the boundaries. It’s not like, okay, you are a startup, so you can lower your compliance requirement to do whatever you want. And the end of the day, you need to be responsible, not just responsible AI for the users, you also need to be responsible for your investors, right? So that’s on the on the startup part. In terms of compliance, think the startups in general do pay a lot of attention to compliance with different, especially now AI, as we mentioned, right? If you look at the APAC, there’s no unified approach to AI governance, right? It’s not like EU has AI efforts. So the compliance cost is indeed very high startups. This is the luxury Big Tech have. We just discussed Microsoft as a case study. So Microsoft has pretty decent size of legal team, security team, enforcement team, to support those three pillars, like safety, security, and privacy. But for a startup company, you can imagine they probably only have one legal, one policy manager for everything.That is it. That is a challenge. And then this is also why a lot of companies complain about very tight regulatory environment in EU because as a startup you don’t want to spend all your money, not even like half of the money as your compliance cost, right? So I always joke with my friends like if you hire more lawyers than engineers for a tech company, I don’t think that’s right. So this is a constant challenge for startup, how to comply, but in the meantime, also keep innovating.Grace Shao (29:28)I think that’s really interesting because essentially, like you said, whether it’s a startup or enterprise, in many ways, it’s faced, they’re facing the same issue. But my issue right now with kind of the AI space is actually there’s lack of international standardization, right? So for example, like globally, wherever you go, you can’t really go stab someone. The rules around drugs or even other issues like driving and other safety issues may vary, but there is like a standardized base normalization or what we believe as humans that should not be done, which is essentially don’t kill people. Homicide is illegal anywhere you go, right?So now with AI, regulation, is that like right now we’re not seeing countries come together and say, this is a sanitized belief that we should just not have. Maybe like you mentioned, child pornography and child safety is something very high on the radar, but even that can be quite subjective from culture to culture. So how do we make of that when we’re going to have AI proliferation across the economy and different touch points in our daily lives?George Chen (30:33)That’s a very important, interesting question and point you make. You you reminded me, you know, when I teach my students in the classroom, one of the examples I give them is, you know, I travel a lot, right? So different countries, to different countries, the first question I ask myself, like, which socket do they use for plug, right? And then even in EU, you know, like, well, in the UK, it’s no longer taught on EU.But even you cross the border sometimes, know, from country to country, you need to, that’s why we always bring a travel adapter, right? So when it comes to AI governance, it’s actually the same problem. You absolutely right in the very spot, EU has the EU AI Act. If you are a startup, you want to expand into EU, no argument, no negotiation, you have to comply with EU AI Act. Plus, several other regulations like the Digital Market Act, the Digital Service Act, then plus GDPR. So to expand into EU is not easy. The compliance cost will be very high. But same thing in APAC. You go to different markets. Indonesia is going to have their own AI regulation versus Singapore, versus other markets. Ideally, the UN should take up a bigger role, a more powerful role to sort of you know, have, you know, control or supervision over like how AI should be used. Right. You know, think about the same question about telephone, you know, when telephone was invented, right? Why, you know, Hong Kong’s, you know, country code is A52, right? Why China is like A6, why US is 001? Because someone made the standard and for telephone code,That was ITU, the International Telecommunications Union. So some people say we also need someone like ITU. Maybe the UN has an AI panel, but I don’t know how powerful the UN AI panel is. I mean, not to mention that the US government is not really a big fan of UN these days. So I agree, we should have international standards on AI, especially on AI safety as a key part of AI governance. We should have some principles.So I think this is something all the countries are looking to. We will very soon have the new annual AI summit in India in February in 2026. I think that India also wants to use the AI summit as opportunity to discuss those standard issues. And also to a point, I don’t know how many already realize, actually US and China are not just competing.In the aspect of AI technologies, like official recognition, deepfake, and other issues, but also compete with each other on AI governance. Basically, the and China are competing, like who’s going to write the rules for AI usage for the next generations. So this is also another flashpoint between the US and China when it comes to technology innovation, not to mention the two countries who are continuing to compete in the aspect of AI technologies, you who’s going to have more faster motors, whether it’s Gemini or DeepSeek in winning the battle. So that would be also a story we watch very closely in terms of competition and struggles.Grace Shao (33:57)Yeah, I think it’s also just because the technology is moving so fast right now. It’s really hard for regulators to keep up even domestically in each country at this point. So yeah, I do agree. I think we need some kind of international standardization. I met with Quaishou’s representative a week ago and it was very interesting to hear. They’re very ⁓ focused on the text to image and text to video kind of space. And basically they said in China,To your point cyber security laws are one of the strictest in the world actually in terms of AI Content AI GC is also one of the strictest in the world She said that actually if you remove the watermark that is actually a criminal offense or like literally you will be like, you know Yeah, it’s quite interesting. And I mean on one hand you think it’s very extreme on the other hand I think it’s very needed right like to make sure that deep fake or the mouth practice or you know, Fabricated content does not spreadGeorge Chen (34:37)That’s right, yeah.Grace Shao (34:52)And kind of lead to the social arrest you mentioned earlier or company disturbance, etc. Or even human to a Harm. Anyway on that note, I want to talk about China. You are interviewed a lot by the media on China US Whether you want to frame as tensions or competition or you know or the race? you know, whatever we want to frame it there is going to be right now two camps essentially, right? ⁓ How do we actually understand the two ecosystems at a high level? Where are the real fault lines? Are they chips, cloud, data, or like you mentioned, regulatory rules? Help us understand the two ecosystems.George Chen (35:21)Right. So let’s talk about China. So US and China don’t just compete in technology. US and China also get more more clashes on AI governance in how the way AI should be regulated. US published the AI action plan under the Trump administration. The AI action plan published by the Trump administration is actually already a shift from the AI policy approach taken by President Biden when he was in office. When Biden was in office, it was more like about protection. Biden focused very much on the online safety and this and that. They even set up the US AI Safety Institute. When ⁓ Trump took over, things changed quite differently. Now, Trump is taking American first approach for know, like America’s version of AI innovation, right? Which means like how we can keep American competitive in the aspect of AI technologies. In the meantime, think the Trump administration also want to export the US governance model on AI to the rest of the world, to many of its allies, especially in Asia, you have like Japan, Taiwan, Korea. While China is also trying to influence perhaps mostly global South countries and Bayer Road countries to be more aligned with China’s AI governance ⁓ model. So the two countries are not just competing in technology, but also in the way how AI should be governed.Grace Shao (37:00)Think on that note, if you’re a developing country right now, whether you’re in Asia, Africa, Middle East, and you’re listening to pitches from both Washington and Beijing, like you said, essentially they want to capture the rest of the world, what questions should you be asking to avoid being locked into one ecosystem?George Chen (37:17)Right, I’m always asked by my friends from global service countries, which side should I take? And my answer is no, you shouldn’t take any side. You should take whatever that fits you to have a sort of combination of the best that you can take from both the US model and also the China model. In some ways, China was quite innovative to solve some unique challenges caused by like a... know, defake and this and that. But in the meantime, the people will say, oh, wow, you know, but you have to sacrifice a lot of, you know, all your privacy, right? You know, even the internet, you you to get on the internet in China, you you must be a real person. know, China has the real ID, you know, policy, right? In Hong Kong, not the same, you you can’t just have, you know, like a, a, a, like a, don’t need to, you know, you have, you know, even for the mobile phone, need to register a number with your real ID. But in the US, this is quite unthinkable. But the real ID approach in Hong Kong and China can certainly strengthen a lot of people’s policy concerns. Versus in the US, everybody can join the party, basically. That will also waste a lot of time. think in China, you will see very much led by companies ⁓ like the traditional BAT and DeepSeek and Huawei to enhance their AI governance through a more company-led plus state-supervised model on AI governance.Grace Shao (38:48)I think that’s really interesting. You just mentioned something that struck a chord with me because I was just in Singapore and I was reflecting on how I basically hated my experience there six years ago, seven years ago when I was single without kids because it feels very like, not control, but everything feels very watched and very sterile and you know, your point, everything is very top down. But this time going as a mother of very young children.George Chen (39:04)Right.Grace Shao (39:10)I loved it. was like, wow, it’s so clean. It’s so safe. I rather give them all my data so they can protect me. They know where to track like ad actors. And I think to your point is very interesting. The idea or the value of Liberty per se may be very different in different cultures and also might change as you go through different phases in your life. So it will be interesting to see how companies or countries choose which ecosystem to join, right? Based on their own.George Chen (39:17)Ha ha haGrace Shao (39:38)belief system or value system. I want to ask you, how are the extra controls right now actually affecting companies operating between China and US? Because I know a lot of your clients probably are operating between these two large economies. You sit in Hong Kong and most of your clients, would assume, are actually like MNCs and have some kind of a... You use Hong Kong as some kind of a gateway to intern exit, mainland China, right?George Chen (39:48)Mm-hmm.That’s right. think that the China model so far, the China model is basically a more multi-barad... the more parallelism, right? So like, you know, to work with different countries, more stakeholders approach. This is also what, know, Premier Li Chang, know, when he was in Shanghai, I think in July for the World AI Conference, you he also called AI as a public goods. You know, I find that that concept was quite interesting. Basically, said this is not just something you and I should exclusively hold, right? This is public goods. This is for, you know, like, almost like, you know, this is like for the fate of the whole mankind in the future. So we need to share the success, share in the growth, versus like the American approach is very clear. You know, this is America first, we need to take the lead. And America has always taken the in AI and technology ⁓ innovation. And again, you know, don’t get me wrong. I think both Li Qiang ⁓ as Premier for China and President Trump have their own very good reasons to manage AI in their own ways. One is to keep raising the American flag high and to make yourself a role model, right? The other is to have a more open model everybody can come and share. I think so far the Chinese model perhaps is more appealing to a lot of developing countries, given the-It’s more like a cost efficiency and a top-down approach also boosts highly efficiency rather than more like a button-up, democratic approach. You need to talk to 10 companies that get alignment, this and that.Grace Shao (41:39)Yeah, actually, you just touched on something I was going to ask you. China has pushed up the AI Plus initiative and they were, like you mentioned, Li Qiang and them are just kind of embracing this idea of exporting AI to global south. But beyond the branding, I was going to ask you, do you think it’s actually successful? But it sounds like they are, right? It sounds like the global south is adopting China’s AI ecosystem because it’s more cost efficient, deployable, scalable given that’s open source, open weight, right? I think I want to ask you one last question on this section, is, you comment on China’s AI ecosystem law in the media. What is something that we’re missing here? What are people kind of missing, maybe even in mainstream media that you think is very important for people to know?George Chen (42:06)Right. That’s an interesting question. think that the international cooperation part is something I’m quite concerned about. China has a lot of good engineers. Actually, we also saw a lot of engineers coming back to China from the US. However, both sides, the US and China, should talk to each other more to maximize the research capacity for the overall interest of the whole mind can. So far it’s not happening. And then the result is, I also think AI, the reason why AI is so special is I believe AI also touches on ideology, the way how people think about things. So the US right now is very ⁓ US-centric, just focused on their AI. And then China is very much oriental focused, trying to focus on their version. So when the two AI is a basically, you know, a developer like in the parallel approach, you know, you don’t talk to each other and that will result in a more divided world when it comes to content moderation, when it comes to understanding, you know, certain issues, you know, which policy approach you take, you know, to explain a historical event, you know, for example, this, if the two countries, US and China, they don’t talk to each other, it’s not going to be helpful.For the overall development in R &D. So again, when I teach my course about the YouTube governance, I use the world’s most popular apps as example. Can you believe, it may not be a surprise for you, actually seven out of 10 most popular apps are in English, originally from the US, very much from California. The other three apps are either from China or Singapore, it tells you something. I think when social media companies began to expand into the Asian region, a lot of countries were fearful of the impact that American social media could bring to their markets. They are also talking about the so-called digital colonialism. Which AI you use that will influence your thinking.So I think in a way, people also need to be mindful whether you are too much into the US model AI, and then that also begin to change the way how you think about things. I actually tested the Chinese AI, and I tried some new features of the AI models. Sorry, I’m trying to think about which model I use. But my point is, the Chinese models are very local, very efficient. For example, when I’m in...Beijing, right? I’m not going to use Chai Chi Bti, not because of BVPN, but I just find like the DeepSeek, you know, the database they have is more practical and efficient and timely than like say Gemini and Chai Chi Bti 4, right? So if I look for the best noodle restaurant, you know, the DeepSeek in Beijing actually, the answer from DeepSeek, you know, could be much better, more accurate, you know, than Gemini and OpenAI.Grace Shao (45:17)That’s really interesting because I think it reminded me of one of the Chinese LLM startups and they said that they’re actually working with local governments in the global south and exactly to your point is that they localize information, localize the culture or language. It goes beyond just the surface language, right? I think that’s really interesting. I wanted to ask you one last question, which is...What is one differentiated view or non-consensus view you hold? This could be about the AI sphere or it could be about something in life.George Chen (45:56)I’m still trying to understand how we should position AI. There is a debate in the industry whether we should position AI as your assistant or more as your partner. I don’t have a clear answer on that. In some cases, I want AI just to be my assistant, which means I tell AI to do what and then you do exactly what I want you to do, right? But...I also understand if you just position AI as your assistant, that that will also limit the potential of the development capacity of AI. But when you treat AI more as your partner, I’m thinking about one of my favorite movies. I don’t know if you remember, there was a movie called Her. There was an engineer talking to the computer. Scarlett Johansson played the sound part.Grace Shao (46:41)Scarlett Johansson, Yeah.George Chen (46:50)for the AIs and that was quite a romantic movie, but the ending was not very, it was not a happy ending. So I’m trying to think like if you produce AI as a partner, you can empower AI to do more things, but then, know, whether eventually we will also enter some dangerous territory, you know, to have AI to have, then we need to talk more like ethical issues, right? You treat the AI more as a partner. There were discussions, know, yes, know, AI doesn’t have feeling.But should we also ask the AI to work for like a nonstop? If you think about human rights, should, if human will work for like 10 hours, 12 hours, you should have to get a break. Why we shouldn’t just keep asking AI all these questions, keep them running the models to get the result. And even AI doesn’t have the touch feeling, does AI have emotional intelligence? I believe at some point that AI will have emotional intelligence. That is to say, if you use AI as sort of a slave.they will also be unhappy. So back to my point, I don’t have a clear answer, but I’m still wondering which sort of status, like a category, we should put AI into, more as AI assistant or as ⁓ a human car parking.Grace Shao (47:58)I think that conversation can like, warrants another conversation on its own, that topic, because I think to your point on the technology aspect, we are seeing a shift from AI just to consumer and just as a chatbot to agent AI, right? So to your point, know, ⁓ AI can actually start completing tasks for you. They can be more proactive, remind you to do things. They are more like a thought partner versus an assistant. But again, to like, you know, to even the conversation we had earlier, it’s like who...George Chen (48:01)Ha ha!Grace Shao (48:26)Who can play God? Who is to say, where is the line, right? And your 10 to 12 hour work ethic thing is very Chinese and American. Definitely in Europe, people are not working 12 hours a day. That is like a, that is a normal work day for Americans and Chinese. But again, yeah.George Chen (48:40)That’s right. You already see the cultural difference here, even for the real world and for the AI in different parts of the world.Grace Shao (48:46)Right? So who gets to set the standards? And I think it will become harder and harder. It’s because then it becomes more philosophical and ethical than just, you know, ⁓ practical, which is right now what we’re talking about AI safety is just like, okay, child pornography is fundamentally wrong. Like, homicide videos is not allowed. Don’t create fake videos of people doing fake things. That is very black and white. We can almost just all universally agree.But when it becomes a cultural, evidently because they’re culture norms, even language norms, societal norms, et cetera, right? Or even each person’s emotional capacity is even different. Then who gets to decide when AI needs to stop, right? That’s definitely like a very interesting topic. And I’ve been having this conversation with friends as well. It’s like, has technology hit a point of actually further development does not progress society as a whole anymore? Or are we still actually benefiting from technological advancements. So anyway, I really, really appreciate that. And I can go on forever. This is an interesting topic. Thank you so much for your time, George. I really, really appreciate your insights and all the expertise and experience you bring to us. AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Why American Investors Should Have a Pulse on China with James Wang 09.12.2025 1ώ 5λ

  In this episode of Differentiated Understanding, I talk with James Wang, general partner at deep-tech fund Creative Ventures, author of What You Need to Know About AI: A Primer on Being Human in an Artificially Intelligent World, and writer of the newsletter Weighty Thoughts. James has sat on nearly every side of the table — Bridgewater investor, startup founder/CTO in healthcare, engineer at Google — and now backs “real-world AI” from semiconductors and interconnects to diagnostics and industrial systems.We start with how the AI investing landscape has evolved since 2016: why “AI” used to be a dirty word in pitch decks, how the post–ChatGPT boom funneled capital into a small set of model companies, and why so many AI startups shot up to tens of millions in ARR only to fall back as incumbents absorbed their features. James explains where he still sees real opportunity — especially in vertical AI built on hard-to-replicate proprietary data — and why moats in healthcare and industrial AI look very different from the “GPT wrapper” era.From there, we zoom out. We compare China vs. the US on AI pragmatism, industrial policy, and consumer vs. enterprise strengths; unpack the open-source vs. closed-source model debate; and talk about how agentic AI is already furthest along in developer tools. James also breaks down the energy reality of AI: why GPUs turn power into intelligence, how much additional load AI really adds to the grid, and what the Inflation Reduction Act and its partial rollback actually changed (and didn’t) for data centers and renewables.We close with James’s differentiated view: that over time, AI’s gains will be largely socialized — diffused into everyday life via cheap, ubiquitous models (often running at the edge) rather than captured as persistent monopoly profits by a tiny set of firms.In today’s world, there’s no shortage of information. Knowledge is abundant, perspectives are everywhere. But true insight doesn’t come from access alone—it comes from differentiated understanding. It’s the ability to piece together scattered signals, cut through the noise and clutter, and form a clear, original perspective on a situation, a trend, a business, or a person. That’s what makes understanding powerful.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.For more information on the podcast series, see here.Topics we covered:* What “real-world AI” means: interconnects, power, semis, diagnostics, industrial systems* How AI investing has changed from “don’t say AI” to “everyone is an AI startup”* Why many high-flying AI startups lacked moats and saw revenue fall back to earth* The case for vertical AI built on scarce, proprietary data (e.g., medical imaging, acoustics)* China’s strength in industrial AI and consumer apps vs. the US edge in enterprise SaaS* Open-source vs. closed-source models, and what really matters for enterprise buyers* What “agentic AI” actually is, and why dev tools are still the most advanced real use case* AI’s power appetite, data centers going “behind the meter,” and the limits of US grid politics* Why James believes most of AI’s value will show up in broad productivity gains, not just in a few mega-capsAI-generated transciptGrace Shao (00:01)Hey everyone, welcome back to another episode of Differential Understanding. Today, joining me is James Wong. James is a general partner at Creative Ventures, spearheading investments in AI across the stack. He was previously the co-founder and CTO of Lioness Health, and before that he was on the core investment team at Bridgewater Associates. He founded a nonprofit consulting firm specializing in microfinance and had a short stint at Google. I’m very excited to actually have you on today, James.James Wang (00:33)Super excited to be here too. Thanks so much, Grace.Grace Shao (00:36)James, it’s really great to actually finally meet you, I guess, in person. We were just kind of laughing about this. We’ve talked on and off on Substack on WhatsApp, on email for quite a while now. I’ve been a really big fan of your writing and you are actually one of the first paid, I think, subscribers to my own newsletter, AI Proem as well. So for listeners, his newsletter is called Weighty Thoughts. He writes everything about the startup space, VC, AI, know, FinTech, I think, and know, bigger pictures as well, right? But ⁓ start with James, why don’t you tell us about your day job? What is it that you do when you say you invest in AI? What are you investing in? And what kind of businesses are you looking at these days?James Wang (01:19)Yeah, sounds great. yeah, glad to be one of the early subscribers to AI Proem because yeah, as a VC, you have to catch the good thing early. it’s a part of the job there. Yeah. So Creative Ventures is an early stage deep tech fund. So deep tech has gone through quite a few evolutions in terms of what it is or isn’t to people.For us, it is things with harder science and IP barriers. So that includes things like battery manufacturing platforms that cost a billion dollars a piece, AI diagnostics that are completely end to end with no clinicians in the loop, things like materials discovery platforms using AI. So a lot of these different areas that have gotten pretty exciting with AI as well. think one of the interesting things is deep tech, especially within say like some of the materials discovery spaces, the bio space, like a lot of these areas have accelerated quite a bit with AI’s ⁓ involvement at this point. And there’s a lot of exciting things coming up in those areas.Grace Shao (02:27)I think you know beyond a business background which a lot of investors have you actually have a technical background as well ⁓ What do you think that like does that make you? More understanding of the deep tech that you’re looking into or do you have any unique perspective on technology companies when you look at investing in them?James Wang (02:47)Yeah, totally. So for us, for our team, actually, I’m one of the few folks without a PhD. So a of the team does actually have that kind of background, which is needed within deep tech ⁓ in large part because it’s you do need to understand how the technology works in order to understand the market that it goes into. That being said, like like most technology startups, the ultimate challenge is finding the right market and scaling.But if you don’t understand the technology on a base level in terms of what it does, it’s really, really hard to actually figure out how to scale the thing. So a lot of that technical background, especially within these areas is quite critical. And I guess just my opinion as well, like a lot of different asset management areas undergo evolution. VC historically has been one that has allowed a lot of generalism within it just because of the nature of how a lot of the software boom went came up and went through and everything. But our opinion is actually a lot of the investors in this particular area will get more and more niche, especially with AI, which I think we might jump into as well. AI does actually involve and help enable a lot of vertically integrated industries ⁓ in interesting ways, which means that, you end up with investors who get more and more specialized in their areas.Grace Shao (04:00)Mm-hmm. How big are these ticket sizes that we’re talking about when you’re investing in and how early are you looking at?James Wang (04:13)Yeah, typically speaking for us, we are often the first institutional investor in that being said, some of our companies have five, 10, even like 20 million dollars in non dilutive government grant funding or research funding before we actually invest. So it’s kind of hard to say when you’re trying to pin that down. That being said, yeah, we’re among the first investors in. Usually we invest around a million in terms of initial check size and sort of ramp from there.Grace Shao (04:28)I see.James Wang (04:42)⁓ And then our companies obviously like as they get larger and larger later on They can have quite a bit of range in terms of like where they end upGrace Shao (04:52)I see. I definitely want to double click on the vertical AI space later. But to start with, another personal question I want to touch on is your book. You’ve just launched a new book. It came out in October, I believe, right? It’s called What You Need to Know About AI, A Primer on Being Human in an Artificially Intelligent World. Can you tell us a bit about the book, a preview of like, I mean, the gist of your book or why we should go read your book?James Wang (05:19)Yeah, totally. Well, think the most recent thing I can remember was someone told me the other day, think yesterday actually, that this is a great stocking stuffer for all the boomers in your life. I believe that was a compliment. So ultimate, I think so. So the book is actually a end to end. Here’s what you need to know to kind of get up to speed on AI.Grace Shao (05:33)It sounds like a compliment.James Wang (05:44)⁓ It goes through the history of it. It goes through some of the technical background. ⁓ Not too deep, but then again, like also doesn’t really pull too many punches in terms of like actually getting into the structure of it. And then finally it goes into how it’s being used today and some implications of it coming up. So it’s meant to take you through end to end. ⁓ You know, we have a lot of interesting endorsements of it. Reid Hoffman actually had read through it and gave a great endorsement of it as well and I’ve had both people within the AI business sector. So basically people trying to market AI and push it out into market as well as engineers, both tell me that they’ve learned something from it. So I think actually a lot of people can get something out of it. It’s just different people will find different parts of the book difficult or not. but it does attempt to like step you through it. so that was the aim that I had writing the book and, ⁓ hopefully I achieved it so far. sounds like it, if it’s a boomer, baby boomer stocking stuffer.Grace Shao (06:44)I think I need to get a book for myself. Does it overlap with what you write about on Weedy Thoughts or is it a bit different?James Wang (06:52)It does, but it ⁓ gets into more depth and basically takes you through A to Z a little bit more. Since I’m sure you know as well, it’s like your experience as well there. It’s like for a substack post, inevitably you leave a lot of things out. You kind of hit the high level, you hit the points, but yeah, ⁓ you can’t make the article 10,000 words long or something like that. On the other hand with a book, you do need to actually bring it from beginning to end.Grace Shao (07:21)What inspired you to write the book? Because I’m sure you have a busy day job already.James Wang (07:24)Yeah, I mean, the first part of it, ⁓ which is the real part of the the marketing story that I give is that which is actually true as well. But the marketing story I give is I sort of looked at the landscape and generally speaking, there’s a lot of great technical resources. There are actually a lot of great sub stack newsletters on AI. A lot of other good places to dive into to get a sense of what’s going on. The problem is in general with the book landscape, a lot of the stuff has been like productivity, get rich quick, et cetera, sort of things within AI, or somewhat more alarmist or very thesis specific driven. It’s like, ⁓ AI is going to do this thing. It’s going to kill us all. It’s going to take all our jobs. It’s going to revolutionize this. Like they’re pushing an agenda. I didn’t really see anything out in the landscape that takes you from A to Z for people who didn’t actually know what was going on and wanted to get up to speed. So I got kind of tired not being able to recommend a book for all the friends who are like, hey, you know what’s going on with the say hi stuff. I don’t know what’s going on with this. I stuff. Where can I go read a single book to find that out? So that was like a big impetus on why I decided to write the book. And in terms of like how I came to it, the publisher actually found me through my sub stack and kept bugging me to write a book. And eventually when I decided to go on this direction, they were like- Are you sure you don’t want to write another thing about how it’s going to kill us all or something? We think that might come off and fly off the shelves a little bit more or like draw people in a little bit more than a light textbook. But nonetheless, it has actually got number one in lot of categories on Amazon for a couple of weeks now. So I think it’s working so far, which I’m happy about.Grace Shao (09:13)I think that’s a really pragmatic way of approaching it. like, your point, I think I’ve also noticed in just the AI sphere in general, there’s like kind of the, you know, the investors are talking about money. What’s the return? What’s the return? It’s only about monetary return, right? The tech people, when you speak to them, sometimes frankly, they’re a bit ignorant to the societal implications or they’re not, most of them, let’s just say, people are not evil intrinsically. They just think, okay, tech acceleration is a fundamental goal for them.And I think sometimes to put the blinders on and they forget about the potential implications of society and the environment on, you know, shifts in like, you know, even dynamics between people and countries. And then like you said, unfortunately, I think the people who really try to bring the awareness and be mindful sometimes can go too extreme. And then what happens is like, let’s reject it. But the reality kind of sits somewhere in between where like you can’t really reject a technology once it’s out of the Pandora box, right?And then you can’t really only look at the value creation in terms of monetary terms. And you can’t really say, OK, let’s only focus on technology, but not think about all the other consequences. So I think to start our interview, let’s really talk about your philosophical view on this. I think your book really ⁓ resonates with me. What I write about also is trying to help people understand, OK, these are the business implications. This is where the return will be. This is what will be.This is how the technology change your interactions with each other. But it’s not like you can’t approach it mindfully, right? So I’ll throw it back to you. How do we understand the relationships between if you have to simplify the three kind of caps that we see right now?James Wang (10:54)Yeah, I mean, in terms of camps, mean, there’s definitely the I mean, it’s been interesting, right? Because for a while you had did have a group that basically said AI is just a fad. It’s not going to actually do anything. Ultimately, it doesn’t matter. Blah, blah. Not a big deal. You have another group that’s basically like, AI is the AI is going to either kill us or AI is going to take off in terms of singularity. And we’ll have a post. ⁓We’ll have post-scarcity utopia where everyone will have UBI and AI will do everything for us kind of thing. ⁓ So between those two extremes, I mean, you do also have people who are like, yeah, this is a great business opportunity. We’ll go after it. It has some aspect of all of these things. And like you’re saying, from a societal perspective, I think a lot of the people who boost AI quite a bit, ⁓Don’t take into account. Yes, there’s going to be disruption. mean, even if you look at the Internet boom, which I think at this point, people have ⁓ misremembered certain aspects of it because it came and went and a lot of our economy has been restructured around it. AI is the same way. It’s going to create disruptions. It’s going to create winners and losers, but it’s going to also help accelerate productivity and do a lot of good in the world, too, like most technologies have ⁓ in the entirety of history.So it’s a big part of just needing to understand what it can do, what it can’t do, and really where we will see the benefits come out. Because I think otherwise, if you’re just very much on the utopian or doomer perspective, you lose the reality of what ⁓ AI actually is, which is a tool, and what the capabilities of that tool is.Actually, in terms of this, I think China currently in terms of Chinese AI, which we’ll probably get into and Chinese AI companies have generally had a more pragmatic view of this. ⁓ I’ve had a lot of conversations in Silicon Valley, including with different folks at the model companies, where some of the goal and some of the ultimate aim was, hey, we’re going to get to AGI and then we win. And then it’s all it’s all done and that like, we don’t have to worry about anything else.I think a lot of that particular mindset viewpoint folly has sort of gone away in the past year or two. But even so, like it tells you something about like the way that a lot of people are looking at this, which is almost pseudo religious.Grace Shao (13:27)Yeah, I think definitely to your point, the the caps kind of become bit of like cults themselves as well. It’s even when you cover this space, it’s interesting to meet people who are like all or nothing, very much all or nothing, right? They’re either like, let’s go all in and ignore all the noise and all the issues or go like, let’s reject this. This is just ⁓ inherently evil, which to your point, like all technology disrupts what we know, butLike you can’t reject it, right? Okay. I think moving on from that, I want to talk about investing in AI. You’ve been investing AI since the early 2020s. I wouldn’t say it’s earliest, but you’ve been in space for longer than most people where, you know, they really jumped in after the chat GPT moment in 2022. So how has that relationship between AI companies and capital change? Like, we’re now hearing a lot of buzzwords. Is this a bubble? Is this like, it’s going to flop? Like, where are you seeing the market kind of likeWhere is it at right now? And has that relationship between the founders and the investors changed over the last few years?James Wang (14:35)Yeah, it’s an interesting question because yeah, when we had started out, and that was 2016, AI was actually kind of a dirty word just because whenever someone tried to throw AI at something, it was like, yeah, this is kind of scammy. It probably isn’t going to actually work. So you literally had different startup pitches pull AI out of their pitch and basically say, no, no, it’s just statistics. Maybe it’s machine learning, but it’s really just statistics and stuff like that.So, mean, the way and the evolution of it like changed quite a bit. I think around 2020 was when it started to get a little bit more accepted. And we started to see like certain pitches where it’s like, hey, look, we’re going to do AI for movie studios. And really what we want is to do motion capture for this or something like this. Actually, I think I saw two or three startups around that period trying to do this. And what you really want to do is like, you really want to fund us. We’re going to gather a ton of data.And then we’re going to train a huge model and then we’re going to win the market. So that was actually kind of a popular thing at that particular point in time, which also ended up becoming unpopular because it ended up not working. but it was post yeah, like some getting some towards the chat, GPT moment that suddenly like everyone knew about AI. took off. A lot of people started putting money into the LLM company, model companies, lot of the companies adjacent to it. And at this point now, I mean, it’s.Interesting, right? Because I think if you look at headlines, would think anything that has AI in its name instantly gets a ton of funding. But I can tell you just being on the ground, that’s actually not hugely the case. Private markets and VC have still been somewhat more sluggish ⁓ since 2022, since interest rates rose, and since there haven’t been super significant exits across the board, which means a lot of that capital market has been frozen.If you look at the stats, actually for all of the startup funding and AI funding, a huge proportion of it is actually just the giant model companies having mega round after mega round, or some of the second tier model companies who’ve also had a bunch of mega rounds. We haven’t actually seen like tons of AI company like Dragon, a ton of startup financing. ⁓ Even so, likeAI currently still is the hot thing. So if you’re trying to raise money, especially in Silicon Valley, you generally will probably try to put some sort of AI pitch into your thing, whether or not it actually makes sense or not.Grace Shao (17:07)Yeah, I was just gonna say it’s really funny when you said like it used to be a dirty word, whereas now you meet any company, like they could be selling chocolate bars and they’d be like, we’re AI, we’re AI company. They’re really trying to use AI as like the kind of buzzword to hook people in. And it’s interesting to hear from your perspective that actually AI startups are not getting a lot of funding, right? Because in China at least, what I cover in this part of the world,There’s already the jokes about the last round of AI startups dying out already, like phasing out recently. Yeah, so I don’t know. Is that happening in SF as well?James Wang (17:45)Yeah, so it’s interesting. maybe I’ll make the amendment that it’s not AI startups are not getting a lot of funding. AI startups are getting more funding than other types of startups. So one of our companies actually just recently was told that, hey, you’re actually an AI startup, not a health care startup. They’re definitely both. But that was their greatest compliment because that meant that the investor was actually interested in putting money into them. So it’s like it tells you.Grace Shao (18:12)That’s so funny.James Wang (18:14)Yeah, it tells you something about the landscape.Grace Shao (18:14)Yeah.James Wang (18:15)⁓ yeah, AI companies get more funding. But yeah, it’s definitely not as much of a bonanza ⁓ as you might think from the outside just by just the numbers thrown at the screen because a lot of the big companies are absorbing. But it is definitely the case that I’ve talked with a couple of other investors who’ve told me about some of the revenue numbers for some of their companies. As much as the revenue numbers shot up, ⁓Grace Shao (18:27)Yeah, yeah.Mm-hmm.James Wang (18:43)Let’s see, I’ll obfuscate what this specific company is, but there’s one company that I know of that basically was like shot up to 20, 30, 50 million in ARR in a very short period of time. And the investor who I was talking to was saying, yeah, they’re definitely going to get to 100, 120, whatever it is. The last time I checked in, I think they had dropped back down to 20 and maybe stabilizing down towards 10. So in terms of AI startups dying, the interesting thing is likeA lot of these startups don’t actually have moats. ⁓ Whether or not it’s like some random model company that likely isn’t ever going to get off the ground in terms of having enough compute resources or whatever, or a GPT wrapper, which has become a dirty word or had become a dirty, like derogatory, like phrase to say to some of these companies that just wrap their product around like a chat, GPT API. A lot of these companies don’t have any barriers to entry.So we’re already seeing them shoot very high up because their products are actually useful. And we can get into that. A lot of these are like programming developer tools, agentic ish tools within developer realm. They go up very quickly. They’re actually quite useful, but then everyone else can utilize, everyone else can build the similar kind of thing very quickly orSay as Codex came out from ChatGBT or as Cloud Code got better and better and added more capabilities, you have a lot of the big model companies themselves end up just incorporating the functionality that these developer tool AI companies tried to put in, but now they’re obsolete.Grace Shao (20:24)Yeah, I think that’s something like I’ve been writing about for the Chinese tech space as well, right? The incumbents end of day have a distribution and what they call the flywheel effect, right? I used to hate the word because I think it sounds really silly, but now I think it really makes sense, right? Like in this case of AI, it’s like if it’s not like we have a new device that we’re interacting with it, so whoever already had existing reach on these operating systems can easily basically just swallow another business, like a newcomer.and just create a function. And to your point on the coding, the agentic side, I know like Alibaba just came out with ⁓ Codar, who I interviewed a while back. ⁓ ByteDance has their similar tool. Obviously Cursor is still the leader globally right now in terms of capability. ultimately, if one day they all reach a similar, I guess, efficiency or user experience, then if you’re already using Alibaba Cloud and you’re already using their like blah, blah, blah service,you’re already buying your API there, then why wouldn’t you just use your tool, right? I’m sure it’s the same in the US, like the big players just kind of capture all at this point. ⁓ I want to talk about creative ventures specifically. You guys say you invest in real world AI, right? ⁓ That’s really much like you kind of even touched on healthcare, robotics, manufacturing. It’s maybe less about like the consumer side of things, right? What are exciting businesses you’re seeing and you think are being overlooked right now?James Wang (21:52)Yeah, I mean, think two areas. One is there are still a lot of interesting things within. For us, real world AI does actually include things like interconnect companies. It includes power management, storage. It includes like different like semiconductor based companies or semiconductor tool companies. So there’s actually a lot of interesting things going on in that realm. It has been a, for example, with like optical interconnects, optical switches, these other things. That’s been aplace where the semiconductor industry has been interested in going for years. And there’s been roadmaps and industry like things talking about like how we need to go that direction. But ultimately, no one actually ever moved because while we have an existing business, it costs time, it costs money to actually move into that. But the interesting thing with the AI boom has been, OK, all of a sudden there is an impetus to actually start adopting a lot of these technologies with some of the optical interconnects and whatnot.And there’s been actually some large exits within the space even recently. So that’s an interesting area to us still. And it’s an area that most investors have still shied away from because there’s still been a historical wariness, especially within VC towards hardware, which is ironic since that’s actually where Silicon Valley started in terms of VC. But on the other side, too, there’s a lot of stuff within health care that has been something that we’ve been pretty excited about, at least in the US. Medtech and health care has gone throughQuite a few years now actually of kind of a funding winter where a lot of well-known health care companies, digital health companies just didn’t do very well. A lot of them also tanked on the public market. So it’s just been a super unpopular area for investors. But some of the most interesting and exciting things that I’ve seen within AI have been within the healthcare sector. There’s some that are basically like healthcare productivity optimization.One of our companies, not to talk up our book too much, but one of our companies is currently the only ⁓ and first and only currently end to end AI diagnostic for them. They’re starting in lung disease, but essentially they’re a diagnostic tool that now you press a button. It sends off the scans. It comes back and gives a diagnostic and gets paid reimbursed by Medicare and all the private insurers. Like there’s no clinician, no technician, nothing in the loop, which actually doesn’t justlike increase margins to software like levels is actually insane in the healthcare sector because there’s just so much red tape. There’s so much red, so many regulatory barriers. There’s so many like pieces that can go wrong and thus you need to check and thus you need to have all these other layers that if a piece of software can actually take all that away and be FDA approved, that’s actually a massive productivity improvement in the space. Again, ours is currently the first and only, but I don’t expect it’s going to stay the only one there’s going to be a lot of really interesting things happening, especially within healthcare and biotech.Grace Shao (24:51)It’s interesting because I was just listening to a podcast with a 16 C’s podcast. They’re saying that actually a lot of biotech and med tech companies are routing their trials actually in China, just given that there’s less red tape around a lot of these issues. And I met with a company in Singapore just last week. They are one of the leading AI companies actually using AI to do clinical research and trial a clinical trials. And it’s really interesting. Like you said, AI is advanced enough in some ways that they can actually guarantee there are no issues in these kind of like more tedious work or knowledge work that it doesn’t really need that much human judgment and then the efficiency gain is crazy. So that’s an interesting space. think you’re right. And I think it’s going to blow up not just in the US, but also maybe in China space as well. ⁓ I want to ask you on China, on China AI, open source, closed source, that’s the big topic, right? It’s a whole China’s embracing open source.The US may be still leading on the closed source models. How do these choices really actually affect the products and the margins ⁓ when you’re looking at these companies? there’s a lot of conversation about their performance, about deployment, but what about when it comes to actual nitty-gritty implementation for the companies? What does it really mean?James Wang (26:11)Yeah, I mean, the way that the Chinese ecosystem and the US ecosystem, it’s partially just path dependent. They’ve evolved in very different ways. In the US, you still have a lot of API or subscription usage, like specifically, you know, open AI and anthropic and Google sticking Gemini in essentially every single productivity tool that they own. ⁓ So the way that that market works is you have a lot of paid usage go out there like they wrap, they do GPT wrappers and whatnot. And because China ⁓ was later to some of the, to some of this in terms of like big breakout, essentially some of the open weight, open source stuff helped ⁓ spur adoption. So for some of the people who do not want to simply pay for chat, GBT or Anthropics API and just wrap their stuff around it and want to control some of their own destiny.It’s great to have something like Quen that you can basically fine tune. can locally host. You can locally host DeepSeq. You might not choose to. Ultimately, you might go to a number of different providers who all allow you to have it available there. But nonetheless, you basically have an easier way of saying that you won’t have lock-in. You’ll be able to use this. You’ll be able to go out there and...Uh, go out there and integrate it. So it is a, again, a little bit of path dependency there. It’s a little bit catching up in, from that perspective as a whole, ultimately in the longterm, I do think like some of the open weight stuff probably does make sense for the same reason. Open source made sense within a lot of the software ecosystem. It does actually spur a lot of enterprise uptake. can pay for support and other things around it. And the bigger thing will ultimately be, can you sell it faster? One of the things about SaaS, so software as a service, has always been, there’s no real barrier to entry for you switching to someone else, except for the fact that I made, especially for enterprise SaaS, an enterprise decision, and I don’t want to switch away from your product now there’s nothing really at the end of the day that makes Salesforce versus some other CRM versus some other productivity tool that different from one another. And at the end of the day, for a lot of the LLMs, especially as we start hitting plateaus in noticeable performance improvement for people. They might become quite interchangeable in which case it becomes similar to a SaaS decision. Do I want to choose the closed source one whereI will have to pay for it forever and also potentially have it go down and only have a single source vendor. Or am I going to take the open weight, open source version where maybe I’ll still pay for it to be hosted, but I can always be comforted that I can always like take it, roll it, and put it in my own infrastructure as well.Grace Shao (29:15)Yeah, I think right now where we’re at as the models, their own performance are getting closer and closer and like the gap is not as wide anymore. I see your point. It becomes like an infrastructure. And then I guess for enterprises, biggest issue or the hurdle is really the switching cost of like the bureaucratic switching costs. It’s like going through the legal work internally, making sure all your departments are upgraded the same way that that becomes a switching cost. So thenI guess incumbents still have the advantage once they’re like chosen as the default provider, they get to kind of own that space, right? ⁓ I want to talk about agents. You kind of mentioned earlier, like enterprises like Google are essentially plugging in Gemini into every single productivity tool. And right now we’re hearing a lot about agentic AI, how that’s going to even increase the capabilities of these productivity tools even further. How do we understandJames Wang (29:53)Yeah.Grace Shao (30:13)what even is a gendered AI right now. I think a lot of people still think of AI as just chat GPT chatbots.James Wang (30:20)Yeah, I mean, agentic AI became a big buzzword. personally, my personal take, I’ll define it a second, but my personal take is eventually agentic as a term will go away. And we’re just going to say AI because all the big model companies are also going that direction in being agentic. Now, agentic, what does it mean? Well, definitions vary, but at the end of the day, it’s, it doesn’t just chat with you.It goes and does something. So instead of I plug into chat, GPT say, Hey, where would be a good place to go on vacation? I would tell the agent, okay, I want to go on vacation. helps me research, but then it also helps me book the tickets, the hotel and like give me like the roadmap and plan and stuff like that and do the things for me. So it’s that layer of being able to interact with the world, whether it’s like true real world or whether it’s like digital world in terms of booking stuff, it can actually do things for you. Now, why, why I say it’s eventually just all going to be agentic, in which case we’re stopped going to stop saying agentic. It’s because the direction of where the model companies have gone is this direction. Uh, as some of the performance differences have disappeared, they’ve implemented more and more agentic tools. would say the most advanced agentic area, even though we don’t usually term it that way, is a lot of the developer tools. Ultimately, at the end of the day, for all the developer tools, they will take your input and they will make changes on your machine, on your code, and commit it. So at the end of the day, it’s doing things.James Wang (32:19)So at the end of the day, in terms of these agentic tools, where developer tools have been the most advanced because they actually go out there and make changes on your machine, your code, commit it, ⁓ all the different model companies have been rolling out more and more tools that specifically are helping you do things in the real world. As you stop having as much difference in how well they chat with you, there’s going to need to be other differentiation. And the place where they’re going, where there’s lower hanging fruit, is being able to actually implement and do things for you. that’s why I think agentic is interesting. Agentic is actually going to be big, but it’s just not that interesting of a term because ultimately most all the AI companies will likely go into it and implement it with their models.Grace Shao (33:07)You’re right, because I think even just this week Deep announced a new like v3 too and they’re saying oh our capabilities are really focused on a genetic AI and every single model is kind of coming out say the same thing and then this is a bit random but it reminded me of this like funny thing was growing up my friend who’s German descent one day asked me she said grace in your household do you guys say eating Chinese food Chinese food do you call Chinese food Chinese food I said no we just say we’re gonna eat food tonight you know and because he normalized and that’s default thing then you wouldn’t really like actually add these prefaced ⁓ adjectives, right? So I get your point about that. ⁓ What kind of agentic products are actually good right now, actually on this point? While models are all becoming more agentic, what actually is it being used in right now? And what tools are actually proving themselves to be really capable and productivity and enforcing?James Wang (34:03)Yeah, I mean, I’ve seen various attempts to do things like ⁓ shopping aids, ⁓ different things with, yes, travel booking, concierge services, ⁓ email responses, bot, chat bot, like customer service ticket, chat bot kind of things. I would still say like among these different things, the most advanced is actually still the developer tools ⁓ ultimately.Grace Shao (34:07)Mm.James Wang (34:30)⁓ it’s close to the companies. It’s close to the people creating it. It’s right now the most advanced area that I see. ⁓ but there’s been a lot of experiments in many of these different areas and they are starting to get better and better and work better and better. So I do actually expect for a lot of these different things where, especially where I guess the framework that I would put on it is if the agentic application is low enough stakes from the perspective of there is a human in the loop that will check it at some point, like, hey, I’m going to book your vacation. Here’s all the pieces of booking your vacation. And you look at it and go, yes, you are correct. I am going to Athens, Greece, not Athens, Georgia, and something like that. If there’s a human in the loop and something where there’s a check, ⁓ these are actually applications that the AI can do quite well and is actually something that’s very implementable currently. So that’s kind of the layer that I put on it. But yeah, like some of these areas are getting quite advanced.Grace Shao (35:32)Yeah, like you said, you need to have that human verification. It reminded me of that new show. It was like a silly rom-com. It’s like they bring these women to Paris, but it’s actually like Paris, Texas, and everyone get off the flight and they’re like, my God, this is not the Paris I imagined. Okay, I wanna kinda go into China a little bit. Like I said, you were one of the first paid subscribers to my newsletter and I very much focus on the China space, although I do cover a bit of APAC and different companies as well. What piqued your interest in China AI? Because it’s not like you actually directly invest in China AI, right? So tell us a bit about that.James Wang (36:11)Yeah, we’re not able to for various ⁓ investment restrictions, some of our investors and whatnot. But at the end of the day, ⁓ it’s a global market. China is a huge market and China also has a lot of talent within it. It’s kind of funny because it’s like, why was I interested, for example, with your newsletter? ⁓ One, you write well, so there’s that. But also it’s just like having the perspective within the market is super important.So I actually still have a lot of conversations with Chinese companies. They know I can’t invest, but they’re always kind of interested in also learning about what’s happening in Silicon Valley, et cetera. So I keep a pulse on the Chinese market that way. And that helps inform my decisions, but also my understanding of like, what does the landscape look like in the U S it’s both compare and contrast, but it’s also thinking longer term. What’s going to happen as all of these companies go more and more global.whether or not they’re competing directly in China or the US, you’re going to encounter each other in the rest of the world, right? So there’s a question in terms of that. As for why, there’s also not very many good sources on China. I think you covered this in some of your articles about some of your own personal history, Grace, but it’s just like, even recently when I was trying to prepare a presentation for talking to some folks about some of the things happening in China where AI is being pushed.especially by the state into a lot of areas like insurance and whatnot. I was trying to Google like what’s going on in China with that, like with just like very simple terms. And really what came up for me was like New York Times articles about Chinese surveillance state is China like doing these things to the Uyghurs is like what it’s like all of these different things that were obviously had its own like bias, let’s say. ⁓ And at the end of the day, it’s like China.Grace Shao (37:36)Hmm.James Wang (38:02)especially for the West has been a little bit of a cipher. It’s either can’t innovate at all and only copies things, or it’s like the crazy, huge country that suddenly like will be able to overtake everyone and like whatever it is. It’s the country that who’s the state dictates everything and thus has like total control and everything. At the same time, it’s like, it’s like a super like, like, you know, lot of the private sector stuff does things, but also like the states can’t seem to like do anything right. And then there’s corruption and ⁓ rockets, rocket fuel being used in hot pot or whatever. And the reality is it’s like, it’s all of these things together. Right. But a lot of the way that the media portrays it is fairly biased in terms of that. So it’s always very useful. And I’d say critical for investors in any part of the world to have a pulse on definitely the two biggest economies in the world. Right.Grace Shao (38:58)Yeah, I think I can go on forever about the media coverage. I wouldn’t even say it’s biased. I would just say the media business model itself actually awards, you know, clicks and attention and in this time and age and what gets attention, the joke amongst a lot of like I expect journalists in the APAC region, it would say it’s big China, bad China, weird China. So I was like, my God, it’s so many people. It’s so big, weird kind of, my god, they eat dogs, which like honestly, majority of people don’t, but sure, I’m sure some certain small segments of people do have weird diets, right? And then it’s like, like bad, bad, bad, like it’s so bad. So I think that gets the clicks, but I do get your point, you know, not to kind of bring it back to myself too much, but it really is why I started writing about it. was like, there should be a nuance understanding what’s happening, especially in the business space when it’s sometimes not really relevant to what we just talked about, the big, bad and weird. It’s just innovation and business. So I want to bring it back to that. lot of people are talking about China being very, very strong and in industrial planning, right? I think this is something that’s all of a sudden for some reason, making headlines all over the U S and last like three months, whereas like industrial planning didn’t come out three months ago. They they’ve been around for the last 30 years, really. From your investor lens, where do you see China moving the fastest? it like only sectors within industrial policy support, like the EVs and the hardware and the robotics? Or do you think China actually has its own mayor and certain sectors are being overlooked? And in kind of comparing that to where you’re seeing the US in terms of the companies you’re investing in, what are things you can actually learn from? I think we can talk about that a bit.James Wang (40:44)Yeah, I mean, in terms of China, there’s definitely a strong advantage in some of those physical industrial areas, ⁓ including say industrial AI, because it doesn’t exist in the US. The US doesn’t really build that many things anymore. It’s really hard to actually get any sort of manufacturing up. I can tell stories in terms of lioness. I can tell stories in terms of some of the med tech companies have helped try to like figure out where to do manufacturing. Essentially, you can’t do it in the US.So all of those sectors and areas ultimately do end up being a very strong advantage for China because it exists there and it doesn’t exist here. In terms of like other things that are overlooked, mean, China and actually Asia in general has had an interesting brilliance within consumer, like the consumer sort of trends there, the apps, the like different ways that, yeah, sure. Like WeChat and other things like.Overall, like China has a much more interesting grasp of like some of the consumer landscape than the US has. I think the, the, I wouldn’t say it’s the, it is, it’s like the stereotype is essentially US companies are the ones that do enterprise SaaS. And the other side of it, which isn’t really spoken at least around here is actually China and actually a lot of Asia is really good at consumer.A larger consumer market, maybe you can argue it’s like some aspects of that, but maybe you can argue it’s some aspect of taste as well. That may be changing ⁓ over time, especially as like the two markets are more and more divorced from each other. ⁓ Ultimately, the U.S. will probably have its own like consumer ecosystem because it’s divorced away from some of the Asian companies in China in particular. China will get separated from like the enterprise sass in the U.S., in which case there has to be its own stack.So maybe that will change over time, but there definitely is like sort of a strong, I wouldn’t say internal cult. I wouldn’t say like cultural from a cultural perspective, but more like cultural from there have been entrepreneurs, successful tech companies and sort of playbooks on like, how do you do this? That are much more mature, say in like the Chinese ecosystem than in the U S ecosystem. Well, you know, the last big consumer app was Snapchat and before that, you know, Instagram and Facebook rewinding to the dinosaurs.Grace Shao (43:06)Yeah, I think that’s interesting. And I think I’ve been hearing more from founders in AIPAC, not just China, but they’re saying with AI, they actually think there’ll be more opportunities in the enterprise space. And the reasoning is because a lot of the reasons why China or South Korea, a lot of these countries at that time, I would say in the 90s, did not want to adopt a lot of SaaS from the US was actually because they frankly didn’t even have the infrastructure in place as in you know internet was not that like you know, you took What’s the word for it ubiquitous and then it was like they don’t have the ability to actually even Serve the you know the need and then on top of that It’s not just China that has a lot of SOEs actually a lot of Asian countries have a lot of state majority companies and I think people again in the West might not realize thatAnd because of that, there’s actually a lot of concern on data privacy issues. And they’d rather have maybe even a shittier quality product than actually jeopardize their data being ⁓ taken by someone else, a third party. So that also goes back to why a lot of companies now in Asia actually want to adopt open source AI versus sending their data across the world. Not quite literally, but giving that re-access to like say a chat jpg or a google gemini so i think there’s a lot of reasons why people might find more use cases of sas ai in asia now ⁓ and they might actually you might see more entrepreneurs in this space popping up ⁓ i think on that i want to ask a question on vertical ai i think we kind of touched on healthcare right now we’re talking about there’s a potential growth in enterprise softwares are ai empoweredHow do we understand vertical AI? Will they still basically have to rely on the incumbents ⁓ infrastructure to build their tools? should we actually kind of see them grow out? should they be compared to the Googles and the Microsofts of the world? Or should they have their own kind of racetrack themselves?James Wang (45:19)I think it’s more their own racetrack because it’s kind of a very different ecosystem based on how AI is evolving. So I think the first layer to talk about is just do any of the LLM model companies have a strong, durable advantage outside of, know, OpenAI and chat GBT has a big brand. Google has a lot of distribution. Alibaba has a lot of distribution. It’s like, is their durable advantage? All of their compute tends to be the same.You know, they’re using the same GPUs. They’re mostly still on CUDA. Maybe it’s moving a little bit, but they’re mostly still on Nvidia GPUs. The models are basically the same. They’re all transformer based models, essentially. And their data at the end of the day, scraped from the internet, is largely the same. Yeah, maybe it varies a little bit, but it’s largely the same. They’re ultimately going to be very, very similar. That just doesn’t give you a lot of differentiation across that. that goes towards like, well, ultimately it might be capped in terms of how different these things are.That goes into the difference between that and the internet, right? The internet was very much an aggregated landscape where it’s like everything’s on the internet. You can go find, like go through to Google, whatever, find whatever you’re looking for. You can go to centralized marketplaces, you know, how about like Amazon, whatever you can find your things. And ultimately like it very much like put people into the same place.What AI is doing, interestingly, is if you think about where there’s actual durable advantage, it’s probably still the computers largely going to be the same. The models in terms of deep learning models, whether they’re transformers or not, are probably also going to be fairly similar. The difference is going to be in the proprietary data. And once you actually get down to the proprietary data level, we’re not just talking about, within your individual corporation, you have your OK.Internal documents or whatever fine. That’s like one thing Maybe you can still use like chat GBT or like one or whatever like some sort of open source LLM But what if you have raw acoustic data from ultrasounds to be able to detect liver disease? You’re not gonna be able to put that into chat GBT at the same time like that data Right now with the current models and compute you don’t actually need that sophisticated of a modelor even that much compute to make it actually do something really interesting with where AI has advanced to at this point. The same has happened with a lot of the drug discovery area, material science area, industrial AI, which again, China has an advantage in terms of this with their actual like running operations and data gathering exercises there. But that’s where the vertical AI comes into play. If you essentially have the ability to have this proprietary data that’s very expensive and difficult to generate,Grace Shao (47:40)Yeah.James Wang (48:09)that you have a great proprietary source for, you can build a durable advantage with your specific models there. So that’s where we’re seeing a lot of split in this area. And the way that this will work is less like aggregation, like the horizontal aggregation of search engines or marketplaces. Instead, what you have is a lot of vertical productivity, where you can build.Like say for drug discovery, can very easily imagine where if you can make a huge amount of productivity increase in drug discovery, that is a massive market, even if you stay just there forever.Grace Shao (48:46)Yeah, and it’s not like these big tech incumbents are going to move into that space. Frankly, it’s not it’s not just like your point. It’s not even just having the money to buy these data. It’s also having the know-how and the like decades of, you know, data gathering that you had to collect. And you can’t really get that immediately right now, right? From the market, from just like the public market or anything. It’s very different from the data you’re scraping from the Internet. Yeah.James Wang (49:09)They also do have their own flywheels that prevent followership. So the flywheel is specifically this. You get the data when it’s cheap because no one thinks it’s valuable. We have one company that literally did this in terms of raw acoustic data. No one usually keeps raw ultrasound data. You usually just have the images, if that, that is kept. No one keeps the raw ultrasound. They bought a lot of it.Afterwards, in terms of other AI companies that come, they go, ⁓ actually, this stuff is super useful. We didn’t realize that they also approach hospitals and say, let’s buy this. The hospitals wise up, right? They’re like, ⁓ this is actually useful. We’re not going to sell you it for next to nothing. We’re going to sell it to you for a lot of money. In the meantime, the other AI company has been using this data to generate revenue, to do these things, to get to a certain level of quality, such that now it’s like, the gap is getting bigger and bigger. The data is getting more and more expensive because it’s becoming like clear that it’s valuable. At the same time, the level of productivity, the level of quality of your AI is likely going to lag the incumbent that already gathered a bunch of data and is generating data at the same time and continuing to improve the model. So they have their own flywheel effects with these as well.Grace Shao (50:24)I see, see. And I think, okay, I don’t know if this is too small of a use case, but I can already imagine, like I just had a baby, right? And then, you know, we went through the private clinic kind of system in Hong Kong where you basically get, they know you have corporate insurance, so they just like make you pay it a ton. And you go in every month, which is absolutely ridiculous. But at same time, when you are a mother, you want to check in and see if there’s anything that needs to be, you know, there’s any flaring issues. Whereas I know in the public sector, if you go to the public system,You don’t get like ultrasounds until you’re what, like three months into your pregnancy, 12 weeks in. You don’t get checked very regularly. It’s usually every trimester maybe once or twice max. ⁓ So yeah, like for something that’s not simple, but as predictable as like a healthy pregnancy, I can imagine if you have an AI telling the mom and you just pay like 20 bucks a month, like everything looks normal, everything looks fine.I can totally imagine that being quite useful for the general mass. if they flare up anything like glaring, that’s something that needs more attention, you can then go into a medical staff’s office for proper checkup.James Wang (51:27)Totally.Right. Well, for one, congratulations. But for two, there’s all there’s so think about it. So one of our companies actually has a product like this, not for the consumer. But ⁓ I’m sure after the baby was born, the doctor did a child hip dysplasia check. So basically checking whether or not the child has this specific condition where if you catch it early, all you need to do is it’sGrace Shao (51:41)Thank you.Mm-hmm.James Wang (52:07)not all you need to do. It’s kind of annoying. You do have to put braces on the child and everything, but it fixes through a few months the problem for the child’s entire life. So it’s very worth it versus having a condition the child’s entire life. One of our companies, ultrasound company has an AI. Basically their end to end product is to do a quick sweep with the ultrasound where it tells you hip dysplasia. Yes, no. So in terms of that,Grace Shao (52:29)Mm-hmm.Yeah, actually, my baby, we had to go through for that and you have to go to like you have to wait like two weeks for the specialist to check you and the specialist you have to wait hours in a clinic like it’s a long tedious process and it’s expensive. So I can imagine this just being lot more affordable and you can actually deploy it to the mass like across like mass market a lot faster.James Wang (52:46)Yes. Well, even with that, it’s just like, again, taking it from like a, I don’t know, evil commercial VC hat, but not really. It’s just like, think about it from a hospital’s perspective. That check is really easy to miss. Like you have checklists, you have all of these things, but you can forget quite easily. That is a huge impact if you forget and it isn’t caught, right? At the same time, it’s like, it’s a doctor that needs to go do the thing. It’s a very valuable resource that needs to stop get sectioned out to go do the thing. If you are a hospital and you are able to basically just have a nurse do a quick sweep and scan and tells you yes, no, and you’ve checked it off the box, you’re probably willing to actually pay a lot for that at the end of the day. And it actually is more beneficial for the consumer too because it is actually doing the thing, super important. It gets done. It’s very accurate.A lot of the vertical AI areas are this way. like, it’s not just productivity increases from the, it’s not just like cost decreases maybe is how I see it. It’s not just the cost goes down. It’s just that the quality of it, the productivity of it, the like accessibility that goes up. And if anything, a lot of cases, the hospitals that whatever the vertical AI case is perfectly happy to actually pay up to their cost, previous cost of the thing.because it’s just so much more reliable, easier, and takes away other workflow concerns. So that’s why a lot of this vertical AI stuff is interesting. Its individual price tends to be actually higher, which is not what you typically think with AI, but it’s just more accurate, easier, smoother, and better for the workflow.Grace Shao (54:25)Yeah. I can see that. Yeah, that’s really interesting. I want to talk a little bit about the big picture policies between China and the US right now. I know you invest across the stack, including some of the infra stuff. When we talked about earlier, you said, look at data centers as well, So help us understand this. Data centers aren’t new.Like, you know, AI needs a lot of energy. AI needs a lot of data centers. How do we understand the relationship between these moving pieces?James Wang (55:11)Yeah, mean, the big thing is AI data centers tend to take a lot more power ⁓ in general for lot of the internet services and other things. ⁓ Because if you’re using GPUs inference, these tend to be much more power hungry components. For internet services, part of the reason why SAS could basically make money with queries that are fractions, fractions of ascent.is because essentially it’s almost free. Like you can use a lot of the way that the internet worked is much more around uptime. So if you actually have, and this may get a little bit technical, if you say have like AWS cloud provider share resources, you’re able to surge up and down your capacity and share it across in terms of virtual instances. actual cost of service, a lot of websites, even massive ones is actually not super high. It’s only high from the perspective of like it may be millions of dollars.But then again, you’re making billions of dollars off of your service that you’re servicing it from. It’s actually not very high. For AI in general, ⁓ its inference costs have been dropping a lot. But even so, with larger models, with needing a ton of memory, with needing a ton of these different things, with GPUs that themselves are both power hungry, but also heat, generate a lot of heat, you basically need to spend the currency of AI is essentially power.Grace Shao (56:21)Mm-hmm.James Wang (56:38)You need to spend power to literally power the GPUs or whatever XPUs, like TPUs, whatever thing you’re doing to run the AI. And you also need to cool it, which also is generally active, which means it’s also power. So all of it boils down to, okay, we need to spend power to be able to do this thing. It’s the closest thing to it is actually like cryptocurrencies in terms of you actually think of the one-to-one translation between power and actually the thing, ⁓ like what the thing does. So.Because of that, the sheer density of power requirements means that usually some of these data centers that are trying to serve AI might exceed the power able to be provided from a local grid that was otherwise serving, just like city, resident, and like normal kind of activity. And you are seeing a lot of these data centers for that reason basically doing their own power purchase agreements.having their own power plants. So they’re not actually on the grid, but they’re basically connected to their own power plants or connected to some of these power systems that are not within like say residential grids or something like that. So that’s been a big part of like why AI has needed that.Grace Shao (57:38)Mm-hmm.As like a average user of AI, should that mean that we should just be more mindful and not use so much AI? Or does that mean that the future of energy consumption will drop as technology advances? Like, how do we understand that? Because like, you know, when we use the internet, it’s not like we think about, my God, how much power consuming, right?James Wang (58:11)Yeah. And the thing that I said before was if you take various stats, it’s somewhere between like 70 to 90 % decrease in inference cost each, like each year. So why haven’t like, you know, inference costs falling through the floor while we’re getting more advanced models, we get reasoning models, which actually use way more tokens or words in order to spit out like the same number of tokens that you see.Grace Shao (58:36)Yeah.James Wang (58:38)So we’re using more and more and more. And that’s why, even though the cost has been dropping so rapidly, we’ve basically kept pace or exceeded it in terms of power. That being said, there’s a question. Where will some of that power requirement ultimately go? How much will be needed? And yeah, will it be the case that we end up just needing exponentially more power? So there’s actually a piece on my sub stack that aa hedge fund buddy of mine, hedge fund friend of mine from Bridgewater wrote, he does a commodity hedge fund now. His point is actually, even if you take very aggressive estimates as for how much power needs will grow for AI, it’s around like a 3.5 % incremental. That 3.5 % is basically the growth rate that we had during the 1950s in terms of the US power grid growing.That can pretty easily be hit by renewables, which have intermittency problems. So you basically need battery storage, which is why we also invest in stationary batteries in that area. Or it can be hit by natural gas, or it can even be hit by just retiring coal plants slower. So actually, a lot of the power needs are not as insurmountable as you might think. And I personally suspect it will ultimately be the case that as we plateau in terms of, hey, this thing likeWe don’t need it to like give us like, has much reasoning anymore. We just needed to book us vacation tickets or something like that. That’ll ultimately level off while the requirements in terms of compute costs, in terms of power costs will keep falling too.Grace Shao (1:00:14)I see, I see. And I think it’s also interesting, so just spoke to David Fishman recently. He’s an energy expert on the China space. And he was saying that, like he kind of mentioned in passing the US side, which is like essentially the US energy kind of, I guess conundrum is more exacerbated becausethe center of living has not increased drastically. So people’s consumption of energy have not actually increased drastically. Whereas in China, over the last two decades, energy consumption has been increasing anyway because of urbanization, because of modernization of maybe your home, the economy as a whole. So there’s been more energy planning in China to actually support that kind of energy increasing demand.And when that AI is now part of the picture, it doesn’t feel like a sudden gap that needs to be filled because you have the renewables, you have small nuclear plants being built out, et cetera. So it’s interesting to hear your perspective that actually the increase in demand, the increasing energy demand is actually not that significant. I think, again, headlines of news articles often really highlight that and really showcase a different picture where sometimes it’s more about like, OK.People are experiencing higher utility bills. The grid cannot actually support local economies or local people’s livelihood anymore. It seems like it’s causing a big issue for the average citizen. ⁓ But yeah, thank you for putting that into perspective.James Wang (1:01:42)It’s totally, well, it’s a self-inflicted issue on the US side. Again, like the US has expanded faster than that at periods in its history. There’s a lot of different energy sources that you can actually use to go after that. It’s just the problem is political in part, like the US has a lot of bureaucracy red tape that’s hard to cut through, in which case it’s hard to build anything economically in the US, which is part of the problem. There’s no nuclear being built.Grace Shao (1:01:47)Mm-hmm.James Wang (1:02:09)So like you’re saying, China is actually building nuclear at a pretty rapid clip. The US is at best unretiring or maybe retiring slowly its existing nuclear capacity. It’s actively retiring its coal capacity, whereas China is what building a new coal. I think it’s one or two new coal plants every week or something like that in terms of the pace. like it’s just a very different kind of environment.But it’s also not because yeah, the U S has no technological ability to go after that. It’s yes. Like you’re saying it’s like we have plateaued and a lot of our energy use. There’s also been a big push towards green renewable energy sources, which especially with the U S grid, low power storage, ⁓ it has its own challenges and can’t actually do the base load for AI. So if we wanted to, the U S could actually pretty quickly solve its problem. The question is, is there the political will and is there the willingness to stomach some of the trade offs for sake?James Wang (1:03:09)higher carbon cost.Grace Shao (1:03:11)Yeah, and I think that’s something David talked about as well in that episode ⁓ where it’s like the trade-off in China is more like, okay, we need more energy so we build more coal, but it doesn’t mean that we stop our renewable. But just because we have renewable doesn’t mean that we stop our coal. The trade-off obviously can be criticized, know, environmental issues, pollution, et cetera. But again, it’s just state level, I guess, mandate or state level priorities a bit different. ⁓So we’re not a political show. We’re going to move on from that. I want to ask you about the Inflation Reduction Act. So this relates to what you just talked about, a lot of the push on renewable energy. then Trump kind of taking it 180 degree on this. So the IRA was introduced in 2022. It tried to make solar and wind more affordable on the grid. How did that actually work out?What does it mean now with the Trump’s one big beautiful bill? Give us a high level explanation what’s happening there.James Wang (1:04:09)⁓ let’s see, data center developers keep getting whiplash in terms of renewables being good and then bad and then maybe not so bad, but not good either. Something like that. I think that’s sort of the quick high level. I mean, so, ⁓ a lot of the incentives, ⁓ were definitely something, things that a lot of data centers, lot of other folks, like hyperscalers tried to take advantage of, ⁓ when the inflation reduction act was more the law of the land before, you know,Some of that got thrown out, big, beautiful bill, et cetera. ⁓ But I mean, the big challenge for the US, though, even just stepping back from that, is regardless of how much legislation you throw at it, it’s just like the CHIPS Act, right? You can throw as much legislation at the CHIPS Act to say, we’re suddenly going to build all our chips in the US now, or something like that. And it’s like, well, ⁓ you’re not spending enough money to do that.And also legislation doesn’t like magically change things unless it specifically hit some of the core problems, which is yeah, the US doesn’t have for chips is like the US doesn’t have enough like labor for this like expertise moved over for sure. It’s ever for the power side. The problem actually goes back to the same thing we just talked about. Transmission interconnects lines old, hard to do, lots of red tape, lots of bureaucracy. It’s hard to build much anywhere.unless you’re building in places that might not actually be super optimal for say like data centers. So, you know, some of the South in terms of Texas or Southwest has been more amenable to some of the data center and like power build out. It’s also hot there. It would really be nice to put it in a colder place. So you have less power needs to cool the thing too. ⁓ The bigger story, I think with all of this, there’s been a lot of legislation that the US keeps throwing out.Grace Shao (1:05:54)Yeah.James Wang (1:06:00)Maybe the bigger story I’d say is just the legislation has done some things around the margins. It has not made like a huge 80 20 change, at least from what I’ve seen. It’s like the same problems, the same ultimate macro problems that plague the U.S. and building stuff. And also it’s aging power grids and interconnect problems between different grids are still the same ones, like regardless of the legislative regime that we’re in.Grace Shao (1:06:24)Yeah.An agent issue with the grid is actually like also just a reflection of like, frankly, the US developed and modernized so much earlier than China. And the grid just by nature is older and therefore the capacity and capability is like weaker because technology advance. Right. I think sometimes people forget about that. Just the reality that China didn’t become China that we know of today until like this decade. And the US has been basically the US that we know of today. The last four decades. Right. ⁓Grace Shao (1:07:33)Then I have one last question for you. I have one last question for you and it’s a question I ask every single guest that comes on the show, which is what is one differentiated view you have? Our show is called Differentiated Understanding. It’s about how you piece together the information you have and how you form a differentiated view, right? So what is something that you think is a bit non-consensus or against what the majority might think?James Wang (1:07:35)Sounds good. Yeah, I mean, I probably would have said it was my view about the vertical AI thing before, because I was talking about that a lot earlier than a lot of other folks, when there was still the talk about foundational models, which still is somewhat talked about. People are really pushing that a little bit less, that foundational models will cover every single use case in existence. And I think there’s been a lot more consensus moved towards that. So maybeThat was a very non-consensus view I had. The consensus has moved more towards. Let’s see, is there any other big non-consensus view right now? ⁓ I think I have one, actually. So another one. So my personal take, because of the way that LLMs have developed and everything, and a lot of the different AI areas have developed, I actually think a lot of the valueof AI from a GDP economy, et cetera, perspective will ultimately be socialized. I don’t mean that as in the government will. Yeah, I don’t mean the government will take it and redistribute it. I don’t mean like something will happen from that perspective or socialism will suddenly take over the US or something like that. What I mean is in terms of economic theory and whatnot, you can either have excess profits be captured by specific corporations and companies.Grace Shao (1:09:03)What does that mean?James Wang (1:09:25)which is frankly as a VC what I’m trying to invest in and basically have essentially monopolistic power, whatever, and base essentially have a lot of rents from society gathered towards the corporation or the company, or you can have a go to labor or you can actually have that value be socialized. Meaning because of competition, because of diffusion of the technology, because it can’t be controlled as much, it just improves society’s lives.and isn’t actually excess captured by any single company. Even though like we have these huge model companies, they’re absorbing a lot of money, all these different things are happening. My personal take is like, they don’t actually have such strong barriers. Do I think OpenAI will go to zero? No, I think they have a pretty strong consumer brand. Do I think Google will go to zero? No, they have a lot of things to like distribute out. There’s a lot of uses for it. The companies will still survive, but they won’t become like essentially like world like consuming companies in the way that some people have talked about AI or talked about AI as in it’s a sector where a couple of large companies will suddenly take over everything. I actually think AI will diffuse within the economy quite a bit where we’ll use it in our everyday lives, but we won’t necessarily need to pay a company a huge amount to do it. For example, in the future, you might have edge models that just run on a very like a fairly powerful inference chip on your smartphone.And you don’t need to pay ChatGPT or anyone else for that. It’s just something that makes your life easier, better. And it’s just there. So that’s one of my takes. I actually think the majority of the value hard to measure as that is will probably be socialized.Grace Shao (1:11:07)That’s really interesting. think that reminds me of something I wrote about recently and I think we engaged online about this as well, which is ⁓the diffusion of AI will in some way look like the diffusion of internet, where it’s not like we just think of four companies as internet companies anymore, but even the tangible real world. Like, you you think about food delivery, you would have never imagined a food delivery company is an internet company. However, it is an internet company these days, whether it’s Food Panda or, you know, like Maytwan or, you know, Seamless in the US, that’s actually like...not a physical world business only, right? And like when you think of a ride hailing, when you think about even like, I don’t know, apartment hunting, whatnot, it’s not limited to just the physical world. Internet companies actually encompasses all these things that we do. It’s just become the infrastructure. So you’re saying AI essentially will just be part of everything we do and it’ll be empowering everything we do. And it won’t just be limited to like the five companies that we think about nowadays. Yeah. Cool.Thank you so much, James. Really, really, really helpful, really insightful conversation. And I really enjoyed talking to you.James Wang (1:12:16)Enjoy talking with you too, this was great, thanks so much, Grace.Grace Shao (1:12:19)Thank you.AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber. Get full access to AI Proem at aiproem.substack.com/subscribe
• 
  Unlocking the Future of Startups and Super Individuals with Bei Zhang 26.11.2025 42λ

  In this episode, I speak with Bei Zhang, VP of Growth at Tanka, about the company’s mission to empower AI-native founders. The conversation covers why persistent, organization-wide memory is the missing ingredient for truly proactive agents, how Tanka stitches together chat, email, calendars, and documents into a single “remembering” teammate, and what agentic work could look like over the next 12 to 18 months. We also take a closer look at the future of founding teams and how agent tools can enable a super-individual way of working without losing control, auditability, or taste.Tanka sits inside a three-layer stack incubated by Shanda Group. EverMind is the AI infrastructure arm that builds a long-form memory orchestration platform. MiroMind is the research lab, built on Qwen models, focused on long-term memory and reasoning. Tanka is the consumer-facing agentic workspace that applies those capabilities to help startup founders run their day-to-day.All three were incubated by the family office of Tianqiao Chen, the Chinese internet entrepreneur and investor behind Shanda.In today’s world, there’s no shortage of information. Knowledge is abundant, perspectives are everywhere. But true insight doesn’t come from access alone—it comes from differentiated understanding. It’s the ability to piece together scattered signals, cut through the noise and clutter, and form a clear, original perspective on a situation, a trend, a business, or a person. That’s what makes understanding powerful.Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.For more information on the podcast series, see here.Topics we covered:* Tanka’s Mission: to empower future AI-native founders to transform their ideas into successful businesses swiftly and efficiently. * The Problem Tanka Aims to Solve: Founders often struggle with information overload, with critical insights scattered across various platforms such as Slack, Google Drive, and numerous AI tools. * How Tanka Works: Tanka’s unique AI memory framework.* The Team: Tanka’s diverse team is rooted in the heart of Silicon Valley, comprising individuals with rich backgrounds in big tech and startups. * Competition is not with general agents—focused and niche market.* Connecting Founders with Investors: It actively seeks to connect founders to investors, creating a community and offering consulting services as well.* Risks of Using AI agents: Human quality control remains essential; a hybrid model is a sustainable, long-term work model.AI-generated transcriptGrace Shao (00:00)Hi, Bei thank you so much for joining us today. I understand you lead Tanka’s growth right now. It’s a very, very exciting startup. I’ve heard a lot about it. Why don’t we start with your role and just tell us about the company, Tanka, the founding mission, what problem you guys are trying to solve, and just a bit about the team.Bei (00:16)Sounds good. Sounds good. Hi, Grace. Thank you for having me. Hi, everyone. My name is Bei. ⁓ I lead the product and growth in Tanka. Before I joined Tanka, I had been in various roles in different AI and SaaS companies, mostly in the GTM function. So what Tanka is about? Tanka is on a mission to empower the future AI native founders to go from ideas to founded very fast, very efficiently. And the core technology we’re putting behind the Tanka is the long-term memory behind the agents. End of the day, we’re trying to create a proactive companion, or we say that AI co-founder, because compared to typical AI chatbots, we are putting more power behind our AI agents that can remember all the conversation, remember all the relationship, and eventually can be the proactive.AI partner to propel the founder to move as fast as possible. So that’s essentially our mission. And we’re hoping we believe the future is the world of super individuals and the lean teams. We’re trying to make the Tanka to be the powerful operating system for the future startups.Grace Shao (01:25)So I think there’s some fun and irony in that, right? what does it mean really when you say it’s an AI co-founder? Like for someone like myself, I’m a independent or I would say like a founder of a startup, I have a small team. What is Tanka really helping me do in a very practical sense?Bei (01:43)Yeah, yeah, great question. We are essentially the kind of startup, so we help ourselves, right? We’re trying to leverage the resources to help others too. what it would mean, maybe we’ll take a step back to get down to the problems we’re trying to solve. Essentially, we being the center of the Silicon Valley, we’ve been hanging out with a lot of founders, or a lot of individual, a lot of lean teams, a lot of them are just like you, Grace. ⁓ You are a super individual. We also have friends being just a three to five person teams. And the common problem we’re seeing they’re facing is the highly scattered information, overload of information, a bloat of different AI tools, and a very spread of key knowledge across different platforms. So even though we’re saying we’re putting the many of the platforms are wonderful. You got all the nice conversations on the Slack. You got all your documents in Notion and the Google Drive. And there are some offline chats. I’m sure there are valuable informations embedded into various GPT tools or AI chatbots. So the core challenge is not having the right tool. The core challenge is when founders are all of a sudden going from a single threat, trying to take on the world, trying to build a business, the tendency is that there is overloading of the information from all kinds of directions. Because for example, we’ve had a very good friend being a very technical researcher in Stanford. But the moment when he or she step into the founder role, he or she will have to handle not only the product, but also engineer the sales, the marketing, the product dev, legal and tax and BD, right? All kinds of stuff going on. So essentially having all those information scattered in different places create a few effects. Number one, it create a huge overload on the human brain, right? Nobody can process the information so effectively. Especially, we even come across multiple founders doing multitasking because they are trying different ideas, right? Which they will just multiply the pins. And separately, when the brain is overloaded, it instantly distracts the founder from the core duty, which is building the product. So that is causing many problems to be happening. It is causing key information getting lost. It is causing one part of the valuable information not necessarily getting fit into the other nice tools or very powerful AI agents, so the outcome isn’t as optimal. It’s far from it, right? The outcome is far from optimal when they’re trying to make a progress on the project. So that’s the mission. That’s what we’re trying to solve in Tanka. So in Tanka, here are a few things we’re trying to tackle the problem. Number one is the AI memory. Without putting the fancy word out here, just thinking...As of you have, let’s say, today, whichever, most of the AI tools are not really memorizing your conversations. Because when you open a window, it has a conversation with you. But the moment you close the session, it doesn’t really record anything. So the next conversation is new. So with the 10Cut AI memory framework, all the conversations and all the documents you put in the tool are automatically compressed, stored properly, and also stored with a high fidelity so that when you have a conversation once, the future conversation will always remember what you had before. So it put a piece of mind to founder’s head so that you know there is a trusted partner that never forgets anything. So every company is about moving forward, not to remember what happened in the past. So on top of that, we’re adding the connectors, making sure Tanka can digest information not only happening within Tanka, but also connected from other sources as a deep memory and context. And with the memory, we’re able to put in the right AI agents, whether to produce the business plan, whether to just do the deep thinking and a deep conversation, or whether to produce an investor-ready pitch deck.They are all based on the actual information in greater details, without you having to chase across all different things. So that’s what we say. That’s the actual specifics we’re putting in behind the tanker, because we’re not calling that just, we want to go beyond the typical AI assistant, because when we say AI assistant, meaning there is some, it’s a reactive, right? There is a AI sitting there and waiting for me to ask the questions or waiting for me to give the proper prompt. So we almost have to treat the typical, even for the very powerful AI chat bot, we have to carefully curate. We have to carefully protect the conversation, making sure it doesn’t generate anything wrong because garbage in, garbage out principle. But with Tanka, because the more you work with Tanka, the more Tanka knows about you, we almost can forget about prompting. It is an actually intelligent person sitting right next to you as a founder. So whenever the conversation happens, we just keep marching forward. And we’re even building more of a proactive AI functions because now that Tanka knows everything, what do we have happening in theory? You should know what I need to do next. even before, in theory, even before I ask,Tanka to do anything, there should be more proactive actions. For example, hey, I need to follow up with certain investors. I need to update the pitch deck, for instance. Some of them are already realized, and many are definitely on the road as we speak. But that’s what we mean by AI co-founder, because we want to essentially have an AI that can essentially propel you to go forward instead of just waiting there for you to comment the way I do things for you.Grace Shao (07:43)That’s super interesting. think to me, when I heard that, I was like, that’s going to be so helpful for me. Cause like you said, there’s so many to do things on the to do list every morning. And then if someone’s actually proactively reminding me or getting things done, that would be really helpful. I first want to talk about the team.First before we get into the product. Just like I understand you guys have a pretty diverse team. A lot of you guys, ⁓ including your founder, came from even ex big tech. How did your team come together? What’s the background? And I guess what is your edge right now making an agentic tool like this, especially with a lot of even the big AI labs are pushing out agentic tools. Like what is your niche and edge?Bei (08:05)Yeah, yeah, great question. So you’re right, we’re a very diverse team. We’re headquartered in Redwood City, California. We do have a global team across different parts of the world. But the core leadership and the product team are right here located in the center of the Silicon Valley because we are a company building for the founders. We want to be where our customers are to shine light on a few other things you covered. When Tanka was born, essentially it was born within a family office that has been actively curating multiple companies and also has been actively investing in hundreds of early stage startups. All the memory problems and all the context switching, all the information overload are very much experienced firsthand, both for the funding members within the family office and also being well observed by the company, right? The family office has been investing and curating in. So it’s a common problem that hasn’t found a solution yet. So that’s where I would say one of the edge is our deep understanding.We’re not an enterprise tool and we’re not so much to a pure consumer tool. We’re living in a breathing in the startup world because the people has been working in the company or surrounding the company has either been advisors, investors, ex-founders of this kind of startups. So we know the problem from a different angles. So that’s number one. And number two is you’re absolutely right, the CEO, Kisson.She came from a Meta, from TikTok. So definitely had a good discipline and a very structured approach from well-formed companies. She also co-founded another company that has a similar form of Tanka. So she brought in tremendous discipline in both the AI agent and from 0 to 1 and from 1 to 100 scale.And I personally come from Grammarly. I happened to have an experienced growing company at a scale and also helped establish the B2B function from the beginning. And other than that, we do have ⁓ members coming from various startups. So we have all been experiencing the problem, first hand, left hand, right? So that gives us a deep understanding on what we want to solve for ourselves.Grace Shao (10:53)But $29 a month is quite steep, let’s be honest, especially if founders are cost-conscious. I want to understand what was the thinking behind that. And again, how does it compare with peers, even more general AI tools like Manus coming out of Singapore right now, obviously, as well as the incumbents that have been integrating AI into their apps like Slack, Salesforce, Microsoft Teams, even Zoom AI companion, right? Like in some capacity, they’re all trying to become a more proactive, I guess, whether you can call it a co-founder or a colleague per se, they’re all trying to be there to be more present to help you actually get things done, right? How do you compete with such an array of competitors, essentially?Bei (11:36)Yeah, yeah, good question. So to your first question about pricing, we put out a pricing more to create ⁓ a sense of familiarity to begin with. So purely on the number, I think it’s a mid-tier. It’s not that high. It’s not that low either. But it’s something people can, our users can correlate to.And if you look at our free tier, we actually have a pretty generous free tier. We have daily bonuses. I think for lot of users to get a feeling, the free tier actually can get a lot done to truly feel the memory behind the agents. And also, separately, we’re paying much less attention on the pricing versus our attention on the value.Because at end of the day, what our users weigh in is how much benefits, how much value they are getting out of the tool. So we’re just putting the pricing as a stake in the ground. We’ve been doubling down on understanding what our users need. They need a collaboration, so we built the AI agents in the chat to empower the team.They need a generative function to turn the conversation into the actual shareable documents. So we did that. We made a very smooth process to go from the chats and the team conversation into the outcomes without you having to reprompt. The users are also looking for more help in the fundraisin, related features just so when they are ready for investor conversation, they can get it funded faster. So we have a whole pipeline of efforts to empower the founders to realize the benefits. in that, our goal is to make everyone feel like the price is a huge bargain. So that’s something we’ve been actively validating. And also separately, to your point, there are it’s an agentic world, right? Everyone, every company, whether the big ones or whether the startups are making various kind of AI agents. We do keep an eye on a lot of the big names, like you mentioned. I do have a lot of admirations to the great tools. But at this point, our belief is that in this age, the AI tool will come out in different formats and different forms.So I like to think of them as inspirations and role models, right? More so than the competitions. If they are doing something similar, right? We would say, how can we fill our own gaps, right? How can we do better than them? But often than not, actually have way more gaps. We think even this big names are not even addressing between on the path, right? Between the ideas to startups getting funded. So we’re hyper focusing on filling the gaps more so than worry about the competition. Because we believe the world is big. The world is big. In the future, everyone will be a builder. Everyone will be a founder. If a user don’t use us, it will not be because of a competition. It will be because we’re not delivering our promise and not creating the value for the users. So that’s where our minds are, mainly.Grace Shao (14:46)So instead of trying to compete on distribution reach right now, you’re really focused on serving a very niche kind of audience, right? And then really just delivering exactly what they need instead of a general mass audience.Bei (14:56)That’s correct. We’re not trying to build a tool for everyone. That’s the job for the big tech. That’s the job for Tech GPT and Cloud. We are in the center of the Silicon Valley. We are hanging out with all the founders who are using all the tools you’re mentioning, but are still struggling in pushing the ideas into tangible business plan. And even for serious entrepreneurs, they are very struggling in getting connecting to the right investors and getting funded very efficiently. So we’re just hyper-focusing on this persona. Because again, we deeply emphasize wisdom because we are them. So if we get this part of the job done, we’ll be very proud of this. We’ll be very proud of our efforts.Grace Shao (15:40)Actually, one thing you just mentioned, how do you connect these founders with investors? What’s the strategy there? Because that’s not a product strategy. Is that just your connection, your network?Bei (15:50)More so than that. So there are multiple approaches. ⁓ number, think about this in a few different approaches. So number one, this is actually interesting challenge because our founder friends are, most of our founder friends are struggling looking for investors and most of our investor friends are still struggling and looking for quality startups, even though they might be in the same room. So that’s still a ⁓ friction. we tackle this in a few different layers. So many of the founders are not effectively connecting to the investors because they’re not ready. They’re not ready. first, we want to make sure Tanka has the capability for them to chat with the team, for them to carry through all the conversations, and making sure all the minute details are reflected in the business plan and the pitch deck so they appear. They are more buttoned up.So that’s where we do the effort in preparing them to be investor ready, because investors are ready in the other end of the room. So that’s low-hanging fruit. And then separately, we are very active in the Bay Area funder communities. ⁓ So if anything, we have no lack of is there is an abundant funder communities here in the valley.And we’ve been actively in the community facilitating the conversation. We’re inviting investors to give advice on how we can build a tool to better empower the founders. We’re doing this in different directions. So in a way, by having a presence in such communities, we’re already acting as a connector between the two parties. And furthermore, what do we do have on the product roadmap. our features like investor database and the investor matching, because that’s a low-hanging fruit. We do want to provide the founders more value by making it very easy for them to see that based on their business plan and the sector, who might be the right person they should be talking to. we are also evaluating the options such as the data room analyzer or the even warm intros because we’re even discussing with the actual human expert fundraising agencies as a potential layer because we do believe this is, AI is not ready to take over the world yet, As awesome as AI can ever be, humans do bring tremendous amount of value. So on a needed basis, there needs to be a human layer on top of the AI workflow.And even if the human layer just evolved for 10 % of the time, we believe that’s where potentially the 90 % of the value may come from. So this is where end of the day we foresee we likely will build ourself into a hybrid solution where 90 % are conducted by the AI or focusing on this path addressing the problems many of the tools are really not addressing specifically.And we’re connecting the human brain, the different part of the party much, much closer in solving this problem. So yeah, does that make sense?Grace Shao (18:58)And you know where else founders should be talking? They should be talking on my podcast because that’s where investors are listening as well and media is listening. And that’s how you get your story out there as well.Bei (19:07)They should. Investors should be listening, too.Grace Shao (19:14)Investors are listening. Actually, my main audience are investors in the US and Europe. I think, you know, interesting founders should be DMing me now. But on a more serious note, I think you just talked about like, agents can do what 90 % of work, you still got to have 10 % of human quality control, right? So end of the day, what are things at least at this point, or the next, say, 12 months, we can delegate agents, what are things that we still really need that human touch or humanity to kind of guardrail, the kind of progression of technology or our workflow or the usage of AI.Bei (19:49)Yeah, we’ve been thinking this day in and day out. So definitely when it’s related to the information gathering, information collecting, the document generation, document refinement, and web scraping. So without saying the features, that basically meaning how you turn from your conversations and inputs, documents, team chats into the pitch deck, into their data room documents, and how to scrape online, how to go to the linking. Those delegatable missions, those missions that tend to be competitive but yet time consuming. If it’s a delegatable, if you can put into a ⁓ SOP or standard operating procedure, we should try our best to let AI to do this as much as possible.However, we do acknowledge that sometimes it takes a lot of judgment in this process because when the funders are so early, would the investors invest into the project or are they investing into the persons? Most likely, earlier they are, the earlier the investors are putting their weight on the persons. But many of the persons’ attributes and experience are not quantifiable.So there are certain things that I cannot build into the AI agents to automate everything. So that’s where we do need a human to better probably connecting with the macro, better putting in the latest reflections, and better just to step in, making sure we’re not misjudging certain startups in either of the directions. And also separately, I would say, we also, Even with all the AI tools out there, we also had very, very top-notch founders who are deeply in the research world. So they just don’t have time. They are very busy. They do want to focus on building their product. Can they learn how to do the whole fundraising business plan or so? They surely can. But it’s more valuable for them to focus on what they do best.That’s where I think sometimes often it just makes sense for the human layer to just step in and take it over. And it could also be entirely 100 % human touch, which could be well suited for the situation. But just wanting to make it possible whether the human touch is 0 % or 10 % or 100%, it is how this startup works. And we should build our product to be seamlessly connected and adapted to the reality here.Grace Shao (22:22)And I think it’s important to kind of note, like, you know, as you mentioned, as we’re all hyping up the AI agents right now, there is some mindfulness to be said to have to about the potential risks, right? So when people are using AI agents, I think this is as an AI agent question as a whole, not just Tanka but who audits the process ensures there are no mistakes, right? When the machines are starting to complete tasks, how do we actually ensure or how do we human ensure that we minimize the mistakes and the risks that they may come with.Bei (22:55)Yeah, it’s increasingly a more critical question as the adoption rate for the AI are increasing. So I don’t have a perfect answer. I don’t think anyone has really found the answer yet. I would say it’s the process. Process meaning when we’re building the product, because we’re building Tanka to be very deep thinking, deep researching, and working on very, very serious projects.We try to use our best model, most expensive one that does the deep thinking and the reasoning to the best extent. So we don’t try to save money using the cheaper model for faster speed, ⁓ which might be introducing more errors. We’re carefully balancing that. We would rather deliver higher quality at a higher cost, but for higher quality. So that’s number one.And number two is because that’s really actually where the memory comes in. Whenever we build a 10-cut AI to help brainstorm with the founders on next steps, we make sure it all ties back to the prior memory or it ties back to the traceable sources. for all the conversation and the generations, there is a link back to where you can point out to.But that being said, it’s not 100%. It’s not like we can disregard any human efforts not to look closely. We still are constantly calibrating, and sometimes errors happen. And that’s even because the LLM, sometimes because the core server, it has variations. Maybe a question from the same LLM vendor may generate different answers.One is more correct than the other one. So I would say it takes both efforts, even though that’s why we do want to emphasize the value of the human, because here’s AI. And we as a human, we still need to be very carefully guarding our own outcome. And then we introduce the human expert to further enhance the quality. I mentioned a lot of fundraising, and we actually have a lot of friends and mentors and advisors from other areas, such as sales and go to market, tax and legal, who are actually ready to engage and looking to find ways to help out the founders. So we’re not building us as a marketplace yet, but essentially we do want to, our vision is we do want to make a Tanka to be the center console where the founders work with Tanka, but also using other tools where it applies. We’re not here to replace anyone.And we would definitely encourage or we may build a bridge between the Tanka with the human experts so that the human and the AI and human harmonically work together to further minimize the hallucination and the errors.Grace Shao (25:40)That’s really interesting. didn’t realize it’s kind of like building up an in-house incubator or like a consultancy, right? Like you have Tanka as your main touch point, and then you expand into your human expertise. Actually on the technicalities, I want to ask what models are you using and how is that decided by the agent? What I put in a prom when I’m using your agent, how does the backend look?Bei (26:00)Yeah, so I’ll say, maybe without disclosing a specific model, we do use a combination of the top tier models. maybe that’s the best way to say it. Using the AI memory, I was too aspect. The AI memory layer is built a little differently. It is called EverMind. It’s actually went open source a few days back. So we built our own prior proprietary and memory layer using a set of the algorithm. And that’s one. And when we build our Tanka AI agents, we do have a router option. We do build a AI. We do have a few preset prompt. Whereas depending on the type of the questions and depending on how the different steps of the agents that can execute, you will automatically pick the best model for the task. So it’s not just one, one deal, right? The kind of large language model will vary. It depends on whether you’re asking to generate a rough idea or whether you’re generating a very buttoned up business plan. So it’s different. And separately, I do want to say because of the memory, that’s where things are a little different, right? So because we do have the AI memory,The large language model is capable of working with ever evolving context and the memory. So even the same question would absolutely yell the different answer the more you engage, the more you evolve with the AI. So I would say the LLM is a commodity. They’re very powerful. They are the necessity. But that’s where at the end of the day, we do think it’s probably going to be safe, whether you’re using Google or OpenAI or Cloud. At some point, it’s going to be indifferentiable, So that’s where, how to make sure it works for you, right? Not for a general purpose. It’s more critical.Grace Shao (27:49)That’s interesting. think that’s what a lot of the AI agents companies been saying as well. Like eventually, you know, the user experience will not, the users will not be able to actually differentiate which model they’re using, but it’s really just on how the interface interacts with the user and if it’s for a specific task. So I kind of want to go in on the product itself. Walk us through the product surface. Like, what is the experience like when I’m a user, I’m a founder, when I go on Tanka what should I expect?Bei (28:16)Yeah, we put in so much sense into the product, but if we, let’s say, we simplify, as a founder, you go into the Tanka, first of all, there is a place you can work with Tanka agent one by one basis. So on this cases, it’s essentially not too crazy different compared to the other AI agents out there, right? You still interact with the agent, you still ask all the questions, right?Further develop your initial idea into a very buttoned up plan and further refining and fine tuning on that. Again, the main differentiation is ⁓ our window never closes. Our window stays always on and never worry about missing any information. So that’s the one. And then let’s say you as a founder, you get an idea from ⁓ a raw impression into something more tangible, you need to work with your team, right? And if today, whether the team is your co-founder, or whether it is your friend or your son-in-law, right, advisor, there needs to be a joint effort because often the wisdom come up in the conversations, right? So that’s why we have a second portion of the tanker to be a chat, right? Whereas we, whoever you invite into the tank to discuss the ideas, to hear the feedbacks, whether positive ones or constructive ones, and whether you both share or you all share any external references. All those conversations are precisely memorized and processed to be the high definition by the AI agent. then whenThat’s essentially where ideally your business plan will evolve from your own work. And with the other AI agents, you would have to reprocess the information. You will have to bring all this conversation into a prompting and making sure, let’s say, that GBT understands what you have talked about. But it was tank up because the AI is sitting there. The AI is sitting there with you in the conversation. After you finish the conversation, after you are aligned,You and your partner or your mentor are aligned on certain solution. Well, you can simply tell the tech to say, go make the next version. In that case, there is no transfer of information. And then there is no loss of communication in between. So that’s the next step, because we see the collaboration being a very core part of the founder. Very few people can pull off the one person team, even for one person, assume, right? You as a super individual, you probably collaborated with many, right? To develop your own business, right? And the last but not least is we are building Tankard to be a very open platform, right? Because this is where we fully acknowledge that everyone will probably use some other tools, whether it’s Slack, right? Whether it is Minos, right? My favorite tool. Again, we’re not trying to compete, right? We’re trying to say, if those critical contacts happen in other platforms, we want to make sure there is a way to bring those contacts into the Tanka so Tanka agent can sync with more deeper memory in mind and thus generate more high quality contents, right? And then the other direction is also true because we actually keep the memory well organized. If at some point the organization or the startup outgrow the Tanka capability, and we are building the MCP to make sure all the memories are exportable to the next tools you’re trying to use. So we are here for the specific purpose. And then there is a beginning point and there is an end point. We’re not trying to do everything. Again, we try to do the best in the part of the problem we’re trying to solve.Grace Shao (31:59)That’s super interesting. I was just going to ask you, where do you think founders can outgrow Tanka? Because you’ve been really focused on saying, helping them out in the very early stages. it’s interesting that you’re quite mindful that eventually, if a company grows to certain size, there is potential that the company or the founder himself might outgrow your app and they will move on to the next agent, next tool. I guess on that note, I kind of want to end on a big picture question, which is,Bei (32:05)So yeah.Grace Shao (32:24)What do you think is the future of work for knowledge workers, especially startup founders, what you’re witnessing in Silicon Valley? I think you alluded to this a little bit, that there are more and more of these called super power or super one-person bands, whatever. But what should we expect? Are we still going to see the kind of startups of couple of people with different technical skills kind of coming together, founding a company, to scaling it?Bei (32:38)Yeah.Grace Shao (32:50)and then becoming a big corporation or are we going to see complete that mode, complete transition revolve.Bei (32:56)Yeah, it’s a loaded question. again, we’ve been very actively thinking along the lines of that too. So here are a few things we believe the future will evolve to. Well, there definitely will be big organizations. That’s just the case. Some businesses are better to be at a bigger scale. Let’s say if you build a robot company, you better be. You need a scale. However, we do see that with all the tools empowering people to go from ideas to the apps very quickly, we definitely see there will be exponentially more super individuals. And when we say individuals, it means either one person or either three to five person. Because eventually, everyone, we do see the traditional roles being very blurred, right? There will no longer be like a PM or front end or back end or marketer, right? Essentially one person likely that’s gonna pick up multiple roles, right? I assume, Grace, you probably were many, many roles at the same time as the owner yourself. I think that’s incredible. And then there will definitely, many of the businesses don’t have to be that big. We do see many companies will probably stay it’s pretty small, right, 5 % or 10%. For instance, Gamma achieved a $2 billion valuation at 50%. That’s incredible. And I think there will be more and more companies like that. So that’s what we are inspired to solve for them. And also, adding one more thing is we do think the future collaboration will be multi to multi, right? That meaning is no longer going to be one person being employed by one company for a long time. Because hey, when everyone can do so many things, if that person has a capacity, why couldn’t he work on multiple projects with multiple teams? That’s also where we are creating the tank to be not constrained by an entity. You don’t have to be the same entity because we fully expect anyone can work with anyone. And we want to embrace that and empower that too. And last but not least, again, there are many good thoughts. I think it’s probably a book worthy if we had more time. So I do think this is where, for the first time, in AI can, in the past, in order to value whether the workforce or organization, whether it’s effective, you kind of have to wait until the quarter end or year end to see the outcome, to see that. Because many of the information are not really recorded. But now, because everyone used so many tools, and also AI has a memory, and AI can understand how things work, I think the efficiency will be exploding. Because the AI is able to catch where the inefficiency, where the blocker is happening. That’s also, again, that’s why we built AI to be not just one-on-one, but to be in the team, just so AI can discover, right? It can observe what’s working, what’s not working, and making sure that the team always work. Whether your own team or whether the cross-functional team is always in optimal status before too late to essentially the performance review happening every second. So that’s also back to the super individuals, right? The super individuals can measure their own success in real time and furthermore be more successful.Grace Shao (36:21)All right, Bei, we’ve had a wonderful conversation. I have one last question for you, which is a question I ask every single guest that comes on my show. What is one differentiated view you have or something unique you believe in about the industry, about the future of tech and AI, or even something just in general in life?Bei (36:37)Let’s see, I have a couple, but I’ll pick one. even as an agent, we’re probably, I think it ties back to our conversation today, right? We are building, I am actively building the AI product, right? So we want to build a co-founder or even a super powered AI solutions. But I do want to acknowledge that the penetration and adoption of the AI in the real world is so, so low. And chasing after a technical advantage, going after, I think ⁓ sometimes it’s just almost a wrong direction for builders to say, let’s make this PowerPoint generation even more smoother or nicer. And while ignoring that, there are massive amount of human workforce are not even closely in leveraging even basic AI to do things. They are struggling with the basic data scraping. They are suffering with the basic information gathering and to be truly embedded in their workflow. This is actually tied back to our chat in the whole fundraising journey. I we’re talking about the most capable, the smartest, the bravest, the most ambitious founder who can build everything. But I mean, why are they still struggling in knowing where to find all the investors? Who is the right investor to work with? Am I ready for the investor conversation? What else do I need to prepare? How good is good enough? And what to anticipate?Why there are so many basic questions that are not solved. Sometimes I think it’s a, I don’t know whether it’s a differentiator. I just want to, we are doing practicing ourselves. Sometimes we try not to be ⁓ bad at in how we can build this tool to be better than the other competitors. But we go back to the basis on what problem are we solving? How is the problem, how people are tackling the problem today and how we can leverage the technology to best solve the problem. Because I definitely observe when we go chase after the technology advancement, we’re going after like 0.1 % improvements. But when we go back to the basic problem resolution, when we look at how the real world is being operated, we’re looking at like 90 % of the problem are not even remotely empowered. So that’s where I’d like to put out here.There is still a long way to go. And there are so many things to be built. So I’m very excited about the journey and all the possibility and all the value we can bring to the community.Grace Shao (39:11)Thank you, Bei. That’s really thoughtful. And I think that you do highlight a point where I think when we’re all so embedded in the tech and AI scene, we assume people are all adapting and adopting it. But to your point, actually, the general mass is really not up to speed with it. And there’s so much work that needs to be done in terms of educating them and actually working together and actually a lot of issues are not solvable by technology, but it still requires that human expertise. So really appreciate that. Thank you so much for your time today. Is there anything else you would like to share with us before we hop off?Bei (39:44)Well, first of all, thank you for the time. I love all the very thoughtful questions. It’s been a pleasure chatting with you and I’m grateful for the opportunity to organize the mind and the share with you and your audience as well. The last thing will be just any recommendation, any suggestions is welcome from you. I I hope this is a...This is the start of the conversation, more so than the end of the conversation. And again, you’ve been a super individual. I want this product to be helpful for you. And also, I would love this product to be helpful for your audience, whether they are investors or they are founders. So I’m just putting, I’m definitely very, very open. We’re a sponge. We’re a sponge. We’re here to take on any suggestions or feedbacks and that’s the only way we can get better and really focus on the right problem to solve and we need everyone’s help. So thank you, thank you, Grace and thank everyone in advance for all the nice thoughts. Get full access to AI Proem at aiproem.substack.com/subscribe